Category: Artificial Intelligence (AI)

What if the US built its future on AI factories? And what if AGI arrives just in time to run them?

In this episode, Paul and Mike break down the White House’s aggressive three-part Action Plan, including its call to build more data centers and ban “woke” AI. They unpack what staggering token usage tells us about the pace of AI development—and how that connects to the rumored, unified GPT-5 model that could reshape everything. Then it’s rapid fire: Nvidia CEO’s advice for college students, the first AI for therapy, AI’s impact on tech jobs and more. 

Listen or watch below—and see below for show notes and the transcript.

Listen Now

Watch the Video

Timestamps

00:00:00 — Intro

00:06:23 — White House AI Action Plan

00:31:55 — How AI Could Upend the World Economy

00:39:37 — GPT-5 Rumors

00:47:52 — AI Is Impacting Tech Jobs

00:53:08 — Advice for College Students

00:59:44 — Instacart CEO About to Take Reins of Big Chunk of OpenAI

01:08:32 — The First AI for Therapy

01:12:31 — AI’s Environmental Impact

01:17:04 — AI Search Summaries Result in Fewer Clicks

01:19:45 — AI Product and Funding Updates

Summary:

White House AI Action Plan

The White House has released its official AI Action Plan, a strategy document that frames artificial intelligence as a global race for “unquestioned and unchallenged” technological dominance.

The plan is built on three pillars. The first, Accelerating Innovation, calls for unleashing the private sector by removing “bureaucratic red tape” and “onerous regulation.” It directs federal agencies to rescind the Biden administration’s AI executive order and revise standards to ensure AI systems are free from what it calls “ideological bias.” The plan also emphasizes supporting American workers with skills training for an AI-driven economy.

The second pillar, Building Infrastructure, is a massive domestic push under the mantra “Build, Baby, Build!” It aims to streamline environmental permitting for data centers, semiconductor factories, and energy projects, while explicitly rejecting “radical climate dogma” to expand the nation’s power grid.

The third pillar, International Diplomacy and Security, focuses on exporting the full American AI tech stack to allies while strengthening export controls to deny adversaries access to advanced chips and manufacturing equipment.

The Action Plan is authored by Michael Kratsios, Assistant to the President for Science and Technology; David Sacks, Special Advisor for AI and Crypto; and Marco Rubio, Secretary of State.

How AI Could Upend the World Economy

What if artificial intelligence doesn’t just disrupt the economy, but detonates it? That’s the provocative question posed in a briefing in this week’s issue of The Economist. 

Unlike past technologies, AGI could automate not just labor, but innovation itself, generating ideas, conducting scientific research, even improving its own design. 

If that happens, the global economy wouldn’t just grow, it could explode, hitting 20 to 30 percent annual growth rates.

But growth at that scale doesn’t necessarily mean prosperity for all. As AI gets cheaper and more capable, wages could shrink, and workers might be priced out entirely. 

Capital—not labor—would capture most of the value, meaning those who own AI or data centers could end up with a staggering share of future wealth. Yet even with these projections, markets aren’t behaving like explosive growth is around the corner. 

GPT-5 Rumors

OpenAI is gearing up to launch GPT-5 as early as August, according to a new report in The Verge.

Sam Altman said recently on X that “we are releasing GPT-5 soon,” and he previewed GPT-5’s abilities in a recent podcast with comedian Theo Von.

He told the host that he let GPT-5 take a stab at a question he didn’t understand, and the model answered it perfectly. He called this a “here it is” moment, and said he “felt useless relative to the AI” because he felt like he should have been able to answer the question.

A post on X, right around the time of the podcast was released, revealed that GPT-5 had been spotted briefly in the wild.

The Verge says the full rollout is expected to include three tiers: a flagship model with integrated o3 reasoning, a lightweight “mini,” and an API-only “nano.”

Critically, GPT-5 consolidates OpenAI’s fragmented model lineup into one unified system, which is a move toward the long-term goal of AGI. 

If AGI is ever formally declared, it could shift OpenAI’s business relationship with Microsoft in profound ways, including revenue rights.

This week’s episode is brought to you by MAICON, our 6th annual Marketing AI Conference, happening in Cleveland, Oct. 14-16. The code POD100 saves $100 on all pass types.

For more information on MAICON and to register for this year’s conference, visit www.MAICON.ai.

This episode is also brought to you by our Academy 3.0 Launch Event.

Join Paul Roetzer, Mike Kaput and the SmarterX team on August 19 at 12pm ET for the launch of AI Academy 3.0 by SmarterX —your gateway to personalized AI learning for professionals and teams. Discover our new on-demand courses, live classes, certifications, and a smarter way to master AI. Register here.

Read the Transcription

Disclaimer: This transcription was written by AI, thanks to Descript, and has not been edited for content. 

[00:00:00] Paul Roetzer: People in power wanna stay in power, and if these models from the five companies that are building the frontier models control the power and the trillions of dollars of value, whoever is in power will abuse them. Like that is human nature. Welcome to the Artificial Intelligence Show, the podcast that helps your business grow smarter by making AI approachable and actionable.

[00:00:22] My name is Paul Roetzer. I’m the founder and CEO of Smarter X and Marketing AI Institute, and I’m your host. Each week I’m joined by my co-host and marketing AI Institute Chief Content Officer Mike Kaput, as we break down all the AI news that matters and give you insights and perspectives that you can use to advance your company and your career.

[00:00:44] Join us as we accelerate AI literacy for all.

[00:00:51] Welcome to episode 159 of the Artificial Intelligence Show. I’m your host, Paul Roetzer, along with my co-host Mike Kaput. We are recording. July [00:01:00] 28th, 11:00 AM Eastern Time, expecting maybe some announcements this week. So timestamp might be relevant here. this episode is brought to us by AI Academy, by Smart Rx.

[00:01:11] we have our 3.0 launch coming up. It, I think I mentioned this last week that there was an announcement, pending and it is gonna happen on August 19th. So we have spent the last nine months or so re-imagining our AI Academy and our AI Mastery membership program, and it is launching on August 19th, we’re actually gonna launch with a collection of new on-demand course series and certifications, a new AI academy live with weekly experiences.

[00:01:41] There, a new Gen AI app series that Mike is taking the lead on creating, which is gonna be weekly, 15 to 20 minute product and or feature reviews. It is a complete reimagination and I, maybe it’s something I’ll tell the full story of kind of how we got here. I’ll, I’ll probably actually honestly tell it on the August 19th webinar.

[00:01:59] I still have to kind [00:02:00] of like build that presentation. I’m actually in the midst of finalizing a couple of the new course series, as we speak, taking an hour off in between doing that to do this podcast. So I’ll probably tell the story of kind of how this came to be and, and what version one and two were.

[00:02:17] If you’re not familiar with AI Academy, we actually launched our AI courses in 2020 in lieu of not being able to have a in-person conference that year. We, we launched online courses, so we’ve been doing this for five years and this is a complete reimagination of it, so I’m really excited to launch it.

[00:02:34] The team has been working incredibly hard behind the scenes. We’ve doubled our staff in the last like 45 to 60 days, in preparation for this launch. We expect to continue to grow that staff and the organization as a result of this. We’re grateful for everyone who’s been a part of AI Academy leading up till now.

[00:02:52] We have, I, I don’t even know. there’s been probably over 2,500 to 3000 people go through AI Academy through [00:03:00] the years. we anticipated pretty dramatic uptick in that number very soon, based on early demand for what we’re launching. So yeah, join us August 19th to hear all about it, the vision, the roadmap, an inside look at everything that’s launching that day.

[00:03:16] Any AI Academy members will have access that day too. a lot of the new stuff that’s coming out. And then we’ll share a little bit of the roadmap for where we’re going from here. One of the big features is the new AI Academy will have business accounts, which previously there, that was not a, a feature of it was a lot of individuals.

[00:03:34] so join us August 19th. We’ll put the link in the show notes. You can all also go to smarter rx.ai, and click on education. And the AI Academy 3.0 launch event is right there. So again, just go to smarter x.ai. Maybe Mike will put that in the header too. The CTA, I think there’s currently like a job openings, header.

[00:03:54] Yeah, we’ll maybe we’ll swap that out and put that there so it’s easy for everyone to find. Great. Alright, so that is AI [00:04:00] Academy launch again, August 19th at noon Eastern time. And then also, Mayon our annual in-person event. This is happening October 14th to the 16th. We’ve had incredible response to this so far.

[00:04:13] I think we are, I don’t know the exact numbers. We had a big week last week. I wanna say we’re trending somewhere between 40 and 50% ahead of ticket sales for 2024. So we had about 1100 people at the 2024 event in Cleveland, and we are definitely trending in the direction of 1500 plus. So thank you to everyone who has registered already.

[00:04:34] It’s, it is like the best place if you are a marketer business leader to meet other forward thinking. Marketers and business leaders. again, it’s happening in Cleveland, October 14th to the 16th. Majority of the agenda is published. I’m working on, finalizing the main stage general sessions as we speak as well.

[00:04:53] I actually, I think three of them we finalized. Three or four of them we actually finalized last week. We won’t be [00:05:00] announcing them probably here for a couple weeks, but few more announcements coming, but you can get AGI a general idea of the amazing speakers and sessions and the workshops. The pre the pre-event workshops on October 15th.

[00:05:12] It’s all live right now. Go to MAICON.ai. That is MAICON.ai and you can use the code pod 100 for $100 off your ticket. So when you’re going through the registration process, make sure to enter the promo code POD 100 for a hundred dollars off. Okay, we had, um. What kind of seemed like a slower week.

[00:05:34] Honestly, at first, like as I was looking through all the links going into the weekend, Mike, it was like, yeah, okay. Nothing too crazy. And then honestly, like, you know, sometimes there’s podcasts I prep for where I start to get really excited to talk about the topics. And there is like three, or, I mean they’re all great this week, but there’s like three or four that ended up becoming probably bigger things to [00:06:00] discuss than I, I initially thought at first glance when I, you know, first put ’em in the sandbox of things to go through this week.

[00:06:06] So we got a lot to talk about. starting with the White House AI action plan. Mike? 

[00:06:12] Mike Kaput: Yeah, Paul, I felt the same way. I kind of was like, ah, okay. Might be a little bit of a slow week. And then once we started getting into them, I was like, wait a second. There’s some really important things going on. And yeah, like you said, the

[00:06:23] White House AI Action Plan

[00:06:23] Mike Kaput: first one is that the White House has released it’s official AI action plan.

[00:06:30] This is a strategy document that frames AI as a global race for unquestioned and unchallenged technological dominance. And basically the way they describe this, is quote, this action plan sets forth clear policy goals for near term execution by the federal government. The action plan’s objective is to articulate policy recommendations that this administration can deliver for the American people to achieve the president’s vision of global AI dominance.

[00:06:58] The AI race is [00:07:00] America’s to Win, and this action plan is our roadmap to victory. So with that in mind, keep, keep that at the forefront while we go through kind of the three policy pillars that they built into this plan. And by they, I mean this is an action plan authored by three kind of key people in the administration.

[00:07:17] Michael Kratsios, who’s an assistant to the President for science and Technology. David Sacks, we’ve talked about before. A special advisor for AI and crypto, and Marco Rubio, secretary of State. This plan is built on three pillars. The first accelerating innovation calls for unleashing the private sector by removing bureaucratic red tape and onerous regulation.

[00:07:40] It directs federal agencies to rescind the Biden administration’s AI executive order and revise standards to ensure AI systems are free from what it calls ideological bias. The plan also emphasizes supporting American workers with skills training for an AI driven economy. The second pillar, building [00:08:00] infrastructure is a massive domestic push under the mantra.

[00:08:04] They literally have this in their build Baby build. It aims to streamline environmental permitting for data centers, semiconductor factories, and energy projects, while explicitly rejecting what they call, quote, radical climate dogma to expand the nation’s power grid. Now, the third pillar is international diplomacy and security.

[00:08:25] This focuses on exporting the full American AI tech stack. To allies while strengthening export controls to deny adversaries access to advanced chips and manufacturing. Now, Paul, there’s a ton to unpack in this. It’s like a 28 page policy brief. A couple things that jumped out to me. I mean, we’ve talked about this a ton of times, but my gosh, like you really can’t read this and expect any consideration for AI’s environmental impact from this administration.

[00:08:54] I mean, literally they say their mantra is Build baby build. There’s a ton of stuff in [00:09:00] there about basically streamlining, which is maybe code for getting rid of or ignoring certain environmental, environmental regulations. I also found some of the commentary around AI’s impact on workers. Interesting.

[00:09:12] There’s some measures to drive overall AI literacy. There’s training for jobs in the trades to support all the data centers and infrastructure. And there’s even some discretionary funding to potentially help rapidly retrain displaced workers. So what did you find noteworthy in here? 

[00:09:30] Paul Roetzer: There was a lot. So the document, you, you can see the whole thing at ai.gov and, and, and view it.

[00:09:36] It basically what it does is it breaks down a bunch of areas and then provides like a one paragraph summary and then recommended policy actions. So I, I’ll kind of go through some of the highlights and then a quick summary of the executive orders that were released to, with this AI action plan.

[00:09:56] So the, I guess the [00:10:00] prelude to wasn’t even the introduction, the prelude comes from, it’s signed by, Donald Trump. So it says, today, a new frontier of scientific discovery lies before us, defined by transformative technologies such as AI breakthroughs in these fields have the potential to reshape the global balance of power, spark entire, early new industries and revolutionize the way we live and work.

[00:10:20] As our global competitors race to exploit these technologies, it is a national security imperative for the United States to achieve and maintain unquestioned and unchallenged global technological dominance. To secure our future, we must harness the full power of American innovation. So my very, very high level take on all of this is comes down to competition mainly with Chinand it’s about national security, the economy and power.

[00:10:44] Now, if you go back to last year, you know, we were talking as a lead up to the election cycle last year that this is what America needed to do. So I’m, I’m kind of all for the fact that we are all in on having a plan for ai. [00:11:00] the devil is sort of in the details and the nuance of, as you were kind of alluding to Mike, what they mean by certain phrases.

[00:11:08] Mm-hmm. And, and if you don’t pay close attention to politics, some of this may just sound all amazing and great and, and all where we should be all for. In reality, I think that we have to understand the nuance of, what this administration believes and, and what they’re doing and, and kind of the direction they’re going and what they’ve told us previously about their thoughts on some of these key issues.

[00:11:29] So, with all that being said, kind of break this down a little bit. So in the introduction it says, the United States is an race to achieve global dominance in ai. Whoever has the largest AI ecosystem will set global AI standards and reap broad economic and military benefits. Just like we won the space race is imperative that the United States and its allies win this race Now that on its own, there would be some debating.

[00:11:51] This isn’t a win or lose thing. This is like this perpetual advancement of a technology. There is no point where you say, okay, we, we won or we didn’t [00:12:00] win. So, you know, again, some of the language you just have to kind of put into context here. It then says, winning the AI race will ru usher in a new golden age of human flourishing, economic competitiveness, and national security for the American people.

[00:12:13] AI will enable Americans to discover new materials, synthesize new chemicals, manufacture new drugs, and develop new methods to harness energy, and industrial revolution. AI will enable radically new forms of education, media and communications and information revolution, and it will enable all together new intellectual achievements, unraveling ancient scrolls once thought on readable.

[00:12:35] That has actually happened. That’s why they’re alluding to it, making breakthroughs in scientific and mathematical theory that is happening right now. We just had last week with the International Math Olympiads, open eye and Google Gold medal there, and creating new kinds of digital and physical art, a renaissance.

[00:12:49] So again, contextually, I don’t disagree with any of this, like this is all what AI is going to enable, and it is nice to see the administration, acknowledging that [00:13:00] and understanding that then says several principles cut across each of these pillars. First, American workers are central to the administration’s AI policy.

[00:13:08] The administration will ensure that our nation’s workers and their families gain from the opportunities created in this technological revolution. I bold faced this part, the AI infrastructure build out will create high paying jobs for American workers. They’re basically referencing the build out of energy and data centers there.

[00:13:24] And the breakthroughs in medicine, manufacturing, and many other fields that AI will make possible, will increase the standard of living for all, all Americans. That is, this is commentary here that is not a given. That is, that is a hope and a vision. I would say at this point. AI will improve the lives of Americans by complimenting their work, not replacing it.

[00:13:44] That is a pipe dream. Mm-hmm. So the administration, and again, this is the context and this is as unpolitical as I can possibly make this, I, I don’t care, Republican or Democrat or something in between, like me and Mike don’t see our job to have a political view at all in any of [00:14:00] this. Like our job is literally just to report what is happening and what the current administration believes and is doing.

[00:14:06] No administration in the United States can admit that jobs are gonna be replaced. Like they, they can’t do that. Like if, if the US government straight up comes out and says, yeah, it’s actually just gonna replace millions of jobs, then they would have an uproar and they would lose the next election cycle.

[00:14:22] So nowhere is this administration going to admit millions of people are gonna be displaced or underemployed. They, they can’t do it. So again, you have to take all of this within the context of who is publishing this and what their goals are for publishing it. And that’s just one area to, you know, really understand.

[00:14:41] So then it gets into the action plan. I, I mentioned, so Mike, you had talked about the three pillars and the way the action plan is organized is within those three pillars. And then I’ll just go through like the quick summary and then the highlights of what each of these areas are. So the first pillar accelerate AI innovation.

[00:14:58] It says America [00:15:00] must have the most powerful AI systems in the world. We must also lead the world in creative and transformative application of those systems ultimately is the uses of technology that create economic growth, new jobs and scientific advancements. America must invent and embrace productivity enhancing AI uses that the world wants to emulate.

[00:15:18] Achieving this requires the federal government to create the conditions where private sector led innovation can flourish. So then within that section, these are sort of, imagine these as the subheads, and then underneath each of these that I’m about to list in bullet point form are policy recommendations.

[00:15:35] So the plan itself doesn’t mandate any of this happening. It is basically recommending how to achieve these desired outcomes. Okay. So, again, we, we are in the accelerate AI innovation. These are the subheads within that section, remove red tape and onerous regulation. We’ve talked about how this, administration hates regulation.

[00:15:59] Ensure that [00:16:00] a, that frontier AI protects free speech in American values. The definition in Mar in America of what is classified as free speech in American values has never been more polarized. so again, we have to understand who is saying this, what, what they define as free speech and American values matters, and not just this administration, the next administration.

[00:16:22] So everything within this, and when I talk about being as unpolitical as possible with this, whatever this administration decides, the next administration gets to build off of those principles. So if the next administration decides America has different values or free speech means something different, understand that that shifts the context of this conversation.

[00:16:45] encourage open source and open weight ai. Enable AI adoption, empower American workers and support next generation manufacturing. Invest in AI enabled science, build world, world-class scientific data sets. Advance the science of ai, [00:17:00] invest, invest in ai, interpretability control and robustness. These are all things we talk about on the podcast all the time.

[00:17:06] Build an AI evaluations ecosystem, accelerate adoption and government drive adoption of AI within the Department of Defense Protect commercial and government AI innovations and combat synthetic media in the legal system. So a couple of these Mike just unpacked. So the enable AI adoption is a critical one.

[00:17:25] Their recommended policy action here, to give you an example of kind of the tone of this document. So what they recommend, one of them is establish regulatory sandboxes or AI centers of excellence around the country where researchers, startups, startups and established enterprises can rapidly deploy and test AI tools while committing to open sharing of data.

[00:17:47] So that’s an example of a policy recommendation. Maybe the most important one, at least, Mike, based on the stuff you and I talk about on the pod all the time. Empower American workers in the age of ai. So what, what does that mean? so [00:18:00] here’s a quick synopsis of some of the policy recommendations. Again, these are not things they’re committed to doing.

[00:18:05] These are recommendations advance a priority set of actions to expand AI literacy and skills development. Continuously evaluate AI’s impact on the labor market and pilot new innovations to rapidly retrain and help workers thrive in an AI driven economy. I couldn’t agree more. That is like right fundamental to everything we talk about.

[00:18:24] So to see the US government saying that is, is good news. the next prioritize AI skill development as a core objective of relevant education workforce funding streams. AGI agreed. Great. issue guidance clarifying that many AI literacy and AI skill development programs may qualify as eligible educational assistance under section 1 32 of the IRS code, given AI’s widespread impact reshaping the tasks and skill.

[00:18:51] So in essence, um. The government should support this. They should provide funding, they should provide tax free reimbursements for AI related training. awesome. [00:19:00] Like I, I hope that happens. Like it’s, and I hope it happens tomorrow. Like I hope, you know, three months from now we’re talking about the forward steps being taken in this one.

[00:19:09] Another one is study AI’s impact on the labor market by using data they already collect on these topics. Specifically the Bureau of Labor Statistics, and the Bureau of Economic Analysis. and the Census Bureau. leverage available discretionary funding where appropriate to fund rapid retraining for individuals impacted by ai.

[00:19:27] AI related job displacement. A hundred percent. Like I’ve thought about doing that ourselves, where we would provide, low cost no cost AI education. We can’t, as a private entity the size we are, do that reasonably. It would probably need to be underwritten in some way by sponsors or something like that.

[00:19:44] but I think you’re gonna see this from the major AI labs and the nonprofits, like everybody’s gonna kind of jump in on this and then pilot new approaches. To workforce challenges created by ai, including retraining needs. the next one was build American infrastructure. This is all [00:20:00] about the grid, you know, increasing energy, building more manufacturing of, semiconductors on site, in the us skilled workforce for the infrastructure, cybersecurity, those sorts of things.

[00:20:11] And then pillar three, is, is, yeah, export ai, to allies and partners. Counter Chinese influence and international government’s bodies strengthen AI compute. So, again, those three pillars. I, I would recommend people go read this stuff and yeah, and understand it a little better, but also understand it.

[00:20:30] It is now just a, here’s what we think we need to do. Now it comes down to actually putting this into, into action. And then a quick synopsis on the three executive orders that the best I could find there was three related to this. So the first is, export of American AI Technologies. What does this one mean?

[00:20:48] I, I won’t get into like breeding the whole thing. it means they don’t want China to win. an interesting side note, Mike, I had sent you this one as sort of like a side not originally intended to be in the podcast, but it [00:21:00] fits so well. I, I figure we probably have to address this. So apparently, Donald Trump didn’t know who Jensen Wong, or Nvidia was up until recently.

[00:21:09] So Nvidia, if, if, if you’re a listener and don’t know, is the largest company in the world, they have a $4.2 trillion market cap. And Jensen Wong is the sixth richest person in the world. So I, I think that the Verge didn’t give the timing of when exactly this happened, but it it appears to be since Trump came into office the second time, so since January of this year.

[00:21:30] And so they wanted to go after some of the big companies and, and apparently Nvidia was on Trump’s list of companies he wanted to break up. Hmm. So Trump told this story himself during the AI action plan. Launch event. So I’ll, I’ll just give a little context here because this matters relative to this idea of AI dominance, and, and the infrastructure side.

[00:21:51] So this is from Trump. Before I learned the facts of life, I said, we’ll break him up. Trump recalled, during his speech about his new AI action plan, he [00:22:00] recounted what seemed to be a conversation between himself and an advisor who he didn’t name, who told him it would be very hard to break up. Nvidia Trump said, why, what percentage of the market do does he have referring to Jensen Wong?

[00:22:13] And the advisor said, sir, he has 100%. And he said, who the hell is he? What’s his name? His name is Jensen Wong of Nvidia. The advisor replied, Trump said, what the hell is Nvidia? I never heard of it before. he said, you don’t know what it is. You don’t want to know about it, sir. Trump said he backed away from breaking up a video after he realized it would be counterproductive.

[00:22:34] this is quote from Trump. I figured we go in and we would sort of break them up a little bit, get them a little competition, and I found out it’s not easy in that business. So I said, suppose that we put together the greatest minds and they work hand in hand for a couple years. The advisor said No, it would take at least 10 years to catch him referring to Wong if he ran Nvidia, totally and competently from now on.

[00:22:56] So Trump said, all right, let’s go onto the next one, meaning let’s go break somebody else up. [00:23:00] And then Jensen Wong got to know Trump, and Trump said, and then I got to know Jensen, and now I see why. So what happened was, in the last few months, Trump, who didn’t know who Nvidia or Jensen Wong was apparently, according to his own, testimony here, realized the significance of Nvidia and that it’s an American company.

[00:23:20] The previous administration had put con export controls into prevent the sale of Nvidia chips to China in the fear that China would catch up to us. And so Jensen Wong went, met with Trump and actually convinced him to remove that export control and allow them to sell chips, maybe not their most powerful chips, maybe a generation or two earlier.

[00:23:41] Mm. Sell those chips into China so that America could dominate and they could make the Chinese reliant on American technology. That’s literally the goal there. So this entire part of the I Action plan, the entire executive order is about creating reliance on American technology and accepting that Nvidia is at the frontier of all of [00:24:00] that, and that penalizing Nvidia would be a bad idea.

[00:24:03] This is why NVIDIA’s stock jumped back up in the last couple weeks. So that’s an interesting executive order. There’s another executive order on, um. Accelerating federal permitting of data center infrastructure. So this is like, like you said, Mike, forget any impact on the environment. If it has to do with energy or data centers, we are building it and we are going to win in that space.

[00:24:23] the interesting thing here, I’ll put a link in the show notes for, this is from last fall. Jensen Wong was talking about, data centers. And he says, AI is now infrastructure. And this infrastructure, just like the internet, just like electricity needs factories. These factories are essentially what we build today.

[00:24:42] So he’s talking about NVIDIA builds data centers, but he actually calls them AI factories. You apply energy to it and it produces something incredibly valuable. And these things are called tokens. So what he’s saying is we build energy, we build data centers, those data centers produce tokens, which, basically are the foundation of intelligence.

[00:24:59] And then an [00:25:00] interesting related quote last week from Demis Asaba of Google DeepMind tweeted. You know what’s cool? A quadrillion tokens. We processed almost one quadri quadrillion tokens last month, meaning June, more than double the amount from May. And that was in a reply to Logan Kilpatrick who said, Google is processing 980 trillion plus monthly tokens across our products up from 480 trillion in May.

[00:25:26] So the basically doubling every month the number of tokens being output by these data centers, which means we, as business users and personal users of AI technology are using it that much more, that it’s now outputting all of these tokens. Even if you don’t understand the concept of tokens, it’s basically the equivalent of, of words that would be, if it was a quadrillion, or let’s say 980 trillion tokens, that’s, that’s about, I don’t know, like, 750 trillion words.

[00:25:57] Like the equivalent of that would, would be roughly what we’re [00:26:00] outputting within these models. And then the last one is the most, um. Probably subjectively bias, like depending on your perspective here. the prevent, this is literally the headline of the fact sheet President Donald Trump prevents woke AI in the federal government.

[00:26:16] And so it says they are prioritizing truthfulness and I ideological neutrality. They talk about unbiased AI principles. They say the the large language model shall be truthful and prioritize historical accuracy, scientific inquiry, and objectivity, and acknowledge uncertain where reliable information is incomplete.

[00:26:35] They say they shall be neutral, nonpartisan tools that do not manipulate responses in favor of ideological dogmas like DEI, and that developers will not intentionally encode partisan or ideological judgments into lms. This is the most absurd of all of them because they’re on record saying they want them to output their I ideals.

[00:26:56] So like this administration. the idea [00:27:00] of neutrality is our view of the world. This is what, this is what Elon Musk is doing with X ai. Like he literally said it. We’re gonna train these things to represent what we believe to be historical truths. So this goes back to the episode 1 58 conversation about who decides truth.

[00:27:15] And again, in a nonpolitical way, like if you think that this administration knows what truth is and they present facts, then like, okay, but that means you won’t believe the next administration. Let’s say it’s a demo, you know, the Democrats come back into power, then you will believe that the Democrats are being untruthful.

[00:27:37] And if the Democrats control what a large language model says, and there, I mean, literally within this executive order, it says, that LLM companies, AI model companies will not be eligible for federal contracts if they don’t adhere to the quote unquote unbiased AI principles determined by a biased government.

[00:27:56] So this is the part, like I just, I don’t [00:28:00] understand and I, again, I, I, I go back to last episode’s conversation. I don’t care who you think knows truth and fact there, the opposite. Administration will always come into power. It it’s inevitable in politics. And so we still arrive back at this idea that someone is the gatekeeper of this.

[00:28:20] Whether it’s this administration and you like this administration or you don’t, or it’s the next administration and you like them or you don’t, they will determine this. And if this executive order that mandates following the unbiased AI principles determined by a biased body of people, I, I don’t, I don’t, I don’t get it.

[00:28:39] Like, and so this is where I, you then worry about like the whole AI action plan and how much of it actually falls within the true principles that it says it will follow, which I believe in. Almost all of them. Like the i action plan is a fundamentally solid plan, right? It’s just, [00:29:00] is it going to be pursued in an objective way or not?

[00:29:03] and I would have the same questions regardless of who is in power. Again, this is all about power and controlling these things. The there is believed these things will drive trillions of dollars of economic impact. We’ll talk about that in the next main topic. People in power wanna stay in power, and if these models from the five companies that are building the frontier models control the power and the trillions of dollars of value, whoever is in power will abuse them.

[00:29:28] Like that is, that is human nature. So I don’t know what it means beyond that, Mike. I don’t, I don’t have a, here’s how we’re gonna make this better kind of ending to this. I just want people to understand this is a very important plan. It is a sound plan. It’s well written. Mm-hmm. People who know AI wrote this plan, whether or not it is pursued to the true benefit of Americans.

[00:29:51] At a small scale and more broadly humanity and society. That’s the to be determined part. 

[00:29:58] Mike Kaput: Yeah, and I like your [00:30:00] point too about showing, it shows where this stuff is going. Whether or not these policies get enacted in the right way, we can make some very reasonably confident bets about the future, right?

[00:30:12] Is that the environmental aspect is not going to be a priority that some type of AI literacy is on the table, but it doesn’t address displacement and that I would be betting pretty heavily on anyone that makes data centers moving forward. 

[00:30:26] Paul Roetzer: Yeah, I think that’s a good synopsis. It is. Everything we’ve been saying needed to happen or was happening it does just sort of validate a lot of that and, and again, for me and Mike, like we spent a lot of time researching this stuff, thinking about this stuff, synthesizing this stuff, and we always want to like know that we’re heading in the right direction, that we’re not misleading our listeners and our pursuit of being as objective as we can be about this stuff.

[00:30:51] Then you get a plan from the government that’s basically like, literally like in print saying everything we’ve been saying. It’s like, okay, good. Like we’re on the right track. [00:31:00] We’re interpreting correctly what is going on. And so, yeah, I think like for us it’s helpful to just see it said, and I do, I think AI literacy, they, they, they’re aware of the jobs impact.

[00:31:11] They don’t wanna acknowledge it, you know, directly transparently, but like they’re pursuing ways to solve for it. they’re be embedding on infrastructure. I don’t know that, that it’s the right play to think of it as a race, that we have to beat China at. And we didn’t, well maybe next week we’ll touch on, but like China came out with their own plan like 48 hours later and they were trying to portray it more as like, Hey, let’s all work together.

[00:31:33] And I think it was meant to be sort of like a, I don’t know, sort of the opposite of the US approach. But again, is it truthful? Is it like actually what it was? Who knows It’s politics. Like right, everybody lies, everybody pursues power regardless of what side of the aisle they’re on. 

[00:31:51] Mike Kaput: All right. Our next big topic this week is about the following question.

[00:31:55] How AI Could Upend the World Economy

[00:31:55] Mike Kaput: What if artificial intelligence doesn’t just disrupt the economy but [00:32:00] actually detonates it? And that’s kind of a provocative question posed in a briefing in this week’s issue of the Economist. So in this briefing, they talk about the fact that unlike past technologies, you know, truly getting to AGI could end up automating not just labor, but innovation itself with AI generating ideas, conducting scientific research, and even improving its own design.

[00:32:23] If that kind of intelligence explosion happens, they posit the economy wouldn’t just grow, it would explode. You’d be hitting things like in some projections, 20 to 30% annual growth rates, which are insane the longer they go on. But as the economist kind of unpacks growth at that scale doesn’t necessarily mean prosperity for all.

[00:32:43] As AI gets cheaper and more capable, we could see wages shrink. Workers might be priced out of the labor market entirely. Capital not labor would capture most of this value, meaning those who own AI or data centers could end up with a staggering [00:33:00] share of the future. Wealth created. Yet, with these kinds of projections, if you start gaming this out, if that happen, markets are not behaving like explosive growth is around the corner.

[00:33:11] So the economists kind of unpacked, well, why is that? On one hand, it’s possible the forecasting models being used by some of the more, optimistic AI labs and economists out there are just wrong. Or maybe just like with AI’s capabilities, everyone’s underestimating how fast things are about to move.

[00:33:32] But as one economist they talked to put it in the report, he said, once you start thinking about the impact of economic growth when it comes to AGI, it’s hard to think about anything else. And I think Paul, that last part really stood out to me here because. When you start thinking creatively about the possible effects of like AGI or even, you know, runaway essentially super intelligence that is improving constantly.

[00:33:57] When you think about how that’s going to affect the global economy, [00:34:00] it just becomes kind of a rabbit hole. And I guess my question for you is, are enough people thinking seriously enough about this? I, 

[00:34:07] Paul Roetzer: I don’t think they are. I mean, we, so we talked about, I was going back on like how many times last year we talked about GDP and mm-hmm.

[00:34:14] Economic impact and episode 1 22 jumped out in particular, when we talked about situational awareness from the Upholded Ashman Brenner, from June, 2024. And that was an episode where we kind of got into this a little bit because that was one of the beliefs within, Ashton Brenner’s situational awareness articles was that we could see economic growth rates of 30% per year, beyond quite possibly multiple doublings a year.

[00:34:39] That was just an asinine thing to most people because again, economists, like I say, it’s never happened. Like you can’t, can’t do that. If you look at historical context, it’s just not something that occurs. And so it’s a hard thing for people to wrap their minds around. And so, you know, it largely just kind of gets ignored, at least by the economist I’ve talked to, like they don’t even [00:35:00] acknowledge this as a possibility.

[00:35:01] So, quick backup, GDP growth, gross domestic product, total monetary value of all finished goods and services that are produced within a country’s borders in a specific time period. It’s usually measured quarterly or annually. I pulled this morning, as of June 27th was the last update. the United States GDP, decreased in an annual annual rate of 0.5% in the first quarter of 2025.

[00:35:26] So January through March, according to third, the third estimate release by the US Bureau of Economic Analysis, which is the authority on this. so the GDP is at about 29 trillion. Give, give, or take. you know, somewhere between 29 and 30 trillion currently, but it shrunk in the first quarter this year.

[00:35:45] So again, for someone to show up and say, yeah, it’s gonna grow 20, 30% annually, it’s like, well, just shrunk 0.5%. How could it possibly grow 20? Or it, it’s like a ridiculous thing to consider. So how does AI impact it? Well, it increases productivity. We can [00:36:00] do more, in the same amount of time. It, in theory, drives innovation and new product development, which maybe creates demand for new products and services.

[00:36:08] it creates industry and sector growth. Potentially it boosts consumer demand through personalization of products and services. Now the question is, will people be working and have the income to, to have that demand? Like that’s an unknown. So we can only create more products and services if there’s money to be spent to, to purchase those products and services.

[00:36:28] So. Yeah, I think that this is an example of why Zuckerberg is spending tens of billions acquiring top AI talent. Why hyperscalers like Google and Microsoft have 80 to a hundred billion dollars CapEx expenditures this year. Google just raised theirs in their earnings call last week. They said they were increasing their CapEx this year.

[00:36:48] Microsoft, I think, has stayed steady at their 80 billion. It’s why OpenAI and Xai are pursuing trillions to build out data centers and energy infrastructure. And it’s why we have an AI action plan from the US government [00:37:00] that prioritizes AI acceleration at the cost of everything else. Because even if the 20 to 30% numbers are unrealistic, even getting to like five to seven to 10% would be transformational for the government, right?

[00:37:14] So if you could, you could do that in a consistent way. And so there’s, there’s literally trillions of dollars to be on locked here. And so the companies that can be at the center of it, which largely are. The AI model companies and the companies that produce the energy and the infrastructure to enable those things, build the AI factories like Nvidia, we’re talking about trillions of dollars in market cap.

[00:37:35] And so spending tens of billions or hundreds of billions is nothing for the, for the opportunity and the missed cost. We’ve talked about this on a past episode. I know, because I think, I don’t remember who we quoted on this, but it was like, it might have been Zuckerberg, it was Satya Nadel, or it might have been, I dunno, it might have been Sam Alman whatever.

[00:37:54] the whole idea of it might not work. We might spend a trillion dollars building all [00:38:00] this out as an individual company and it might not work. But what’s the alternative, right? We sit on the sidelines and do nothing, and we’re not part of the conversation. So this is why Meta and Zuckerberg has to be a part of this conversation.

[00:38:11] It might not work, but the alternative is they do nothing and they’re irrelevant in 3, 5, 3 to five years. So all of this opportunity, this possibility of massive growth. Is in large part what is driving all of the investments, all of the actions that we talk about every week on this podcast. 

[00:38:28] Mike Kaput: Yeah. And to that last point, if you are routinely scratching your head or scoffing at the fact, people are investing so much money in AI companies, some of whom do not turn a profit or like cash on fire, this is why it’s a very logical move.

[00:38:43] It’s not stupidity. It may be optimism or mania, but it is not ending anytime soon. Everyone has to do this. 

[00:38:52] Paul Roetzer: Yep. Yeah. If you have the money, and this is why like last week I said there’s basically five companies that can pursue the biggest models. ’cause we are, [00:39:00] we’re talking about hundreds of billions and not to distant future trillions.

[00:39:03] Likes Sam moment kind of came out jokingly last year that he was pursuing 7 trillion. Mm-hmm. I don’t think it was a joke. Like I, I, I don’t know that the number was 7 trillion, but they raised a half a trillion already. Or you know, that’s what Project Stargate is supposed to be. And I can promise you that was just one phase of the grander vision.

[00:39:21] So I am sure that they are at least discussing trillions as what it’s gonna take over the next four to five years to build the infrastructure needed to build the models they envision. Mm-hmm. To unlock all this growth. 

[00:39:35] Mike Kaput: Alright, our third big topic this week, 

[00:39:37] GPT-5 Rumors

[00:39:37] Mike Kaput: OpenAI is gearing up to launch GPT five as early as August, according to a new report with some rumors in the verge.

[00:39:46] Sam Altman said recently on X as well, that quote, we are releasing GPT five soon and he previewed recently GPT five’s abilities in a recent podcast with the comedian Theo Vaughn. And he told the host on that podcast that he [00:40:00] let GPT five take a stab at a question he didn’t understand, saying quote, I put it in the model, this is GPT five and they answered it perfectly.

[00:40:09] He called this kind of a quote here it is moment and said he quote, felt useless relative to the AI because he felt like he should have been able to answer this question right around the same time a post on X revealed that GPT five had been spotted briefly in the while. The verge says the full rollout is expected to include three tiers.

[00:40:29] There’s a flagship model with integrated O three reasoning, a lightweight mini model, and an API only nano model. it’s assumed that GPT five could consolidate open AI’s kind of fragmented model lineup into one unified system and still kind of unclear what that looks like. But that could be a mute move towards this long-term goal OpenAI has of AGI.

[00:40:52] And obviously if we declare AGI at any point, it could shift open AI’s business relationship with Microsoft as well. [00:41:00] So Paul, if the rumors were true, we’re getting GPT five very soon. The unified system thing, we. heard about, known about. I’m not sure if that means the system, system itself will determine which model to use for tasks.

[00:41:13] Like what else is worth getting ready for here if you’re kind of a business leader or a user getting ready for GPT five? 

[00:41:21] Paul Roetzer: Yeah, I think just paying attention to what, you know, OpenAI is talking about when it does come out. You know, understanding the impact. It’s hard to know until we know if it’s a unified model or a router model.

[00:41:31] I don’t know if that’s gonna make a difference, but yeah, I think we discussed the DI distinction there is when you put the prompt in, it may be multiple models still, there may still be a chat model, reasoning model, you know, an image model, and it just automatically decides which model to route it to versus it’s actually just a single model with all of those capabilities built in.

[00:41:50] again, I don’t know, as the user, there might be some latency issues. It might be a little slower if it’s a router model, but I think it’s still gonna do the same, you know, things generally [00:42:00] speaking. The other anecdotal piece is there was, um. There was some rumors that the models were being tested in the LM Arena.

[00:42:07] So they under code names like Zenith Summit Lobster, nectarine, starfish, and oh three Alpha, which wouldn’t be too much. I mean, that, that’s pretty obvious what that one would be. so those have been gotten pulled as of last night. I think they were no longer in the arena. I don’t know how long they were active, but it appeared they were testing some new models that people had pretty, positive responses to.

[00:42:28] My general feeling as I’ve, I’ve kind of mentioned a couple times recently is I, I think we’re kind of at AGI roughly. yeah, that, you know, I think OpenAI probably believes GT five is or will be AGI they’re, they’re kind of alluding to that. It would explain part of their, shift to the talk of super intelligence.

[00:42:46] so I don’t think, yeah, I don’t think that they’re gonna call it that per se. I think they’ll, they’ll probably do a lot of cutesy tweets of like feeling the AGI and things like that. But I just feel like if you take these models and whatever gbd [00:43:00] five is gonna be, and you post train them on some specific things or give them agentic ability to take actions, it likely would qualify for any reasonably historic historical definition of AGI.

[00:43:11] Like, I, I don’t, right. So again, I think it’s just semantics at this point, whether it is or isn’t, it’s hard to really measure. A couple other things that Altman said on the Theo v, podcast that I thought were noteworthy. He said, G pti, GT five is the smartest thing smarter than us in almost every way.

[00:43:30] Meaning is the smartest thing in the room was kind of the perspective here. You know? And, and, and yet here we are. So this is Sam Altman talking to Theo. so there’s like the, it’s so hard to read Sam’s quotes. Sometimes. There’s something about the way the world works. There’s something about, this doesn’t mean it’s true forever, but there’s something about what humans can do today that is so different.

[00:43:51] There’s also something about what humans care about. today that is so different than ai and I don’t think the simplistic thing quite works now, again, by the [00:44:00] time it’s a million times smarter than us, who knows? So he is basically saying G PT five is smarter than him. It’s smarter than anybody else in the room, but yet he’s still there as the CEO of OpenAI doing his job every day.

[00:44:10] You and I are still here doing the podcast. And so like, there’s something unique about what humans bring to the table. He can’t put his finger on it, but like it’s, humans still seem to be needed, even though this thing’s probably AGI based on his own previous definitions of it. he just doesn’t know if that holds true, you know, three years, five years from now.

[00:44:29] And then the other one that had me, had me really thinking, I thought this was a really interesting analogy. He gave, I guess on Joe Rogan’s podcast, Altman had mentioned something about eventually having an AI president. And so Theo V asked him like, Hey, do you think that’s actually like a thing? And so Sam said, hadn’t really taken my thinking to this extent.

[00:44:48] Everything that it takes to be a president, but I know what it takes a lot, takes a lot. People are willing to, man, I really struggled to read his quotes. so, okay, I’ll just summarize this part ’cause it doesn’t [00:45:00] make any sense. he’s basically saying, I know what it takes to be the CEO of OpenAI and so I can better evaluate this on being a CEO versus being the president.

[00:45:09] okay, so CEO, because I know what that job is like. Okay. That should be possible someday. Maybe not even that far. Like, I think the idea to look at an organization to make really good decisions, there’s a lot of things that you can imagine that an AI C-E-O-O-O of OpenAI could do that I can’t do, meaning Sam Altman can’t do.

[00:45:30] And I can’t talk to every person at OpenAI every day. I can’t talk to every user of ChatGPT every day. I can’t synthesize all that information, even if I could. But an AI CEO could do, that and it would have better information, more context. It could massively paralyze this. and I think that would lead to better decisions in many cases.

[00:45:52] So that just got me thinking. I was like, oh my God. He is right. Like imagine if every morning you could do like a one question [00:46:00] poll of your workforce and then like get all that feedback back and like synthesize it in five seconds. A CEO could never do that. Like a human CEO could never do that. And imagine that with your employees, your customers, your board, your, your analytics data.

[00:46:15] Like imagine having real time intelligence and synthesis of that information on any, any data point you want as a CEO. And it’s like, wow, okay. Like that. In that example, you can now start to see where a co CEO that is an AI truly starts to take a greater role in the leading of companies. And then you could apply that to basically any role and say, well, what data do I need?

[00:46:40] Right? What are the KPIs I’m looking at every day? What’s the data I would love to have that I don’t have? what’s the data I have that I can’t possibly synthesize every day and find meaning in, find insights from, make decisions based on, and imagine a generative AI model had access to all of that and could synthesize it into three point bullet [00:47:00] points at any given moment.

[00:47:01] It’s like, whew. Yeah, I hadn’t, I haven’t really thought about it that way. 

[00:47:04] Mike Kaput: Yeah. That would be quite the game changer. I also, as we’re talking about this wonder as well, depending on how GPT five looks, how it uses different models, I wonder if it could be a wake up moment for your average person, because not only being smarter, but I feel like right now people are not already understanding the full capabilities of reasoning models, for instance, because people half the time aren’t even picking the models they’re supposed to be using.

[00:47:30] Yeah. Or picking incorrectly which models they should be using. 

[00:47:34] Paul Roetzer: Yeah. I, I agree. Like if you ask a harder questions and it requires like deeper thinking and you don’t know to go to the O three model. Right. But then the new model. Does that for you and it’s like, whoa, that’s different. 

[00:47:45] Mike Kaput: Yeah. 

[00:47:45] Paul Roetzer: Yeah. I could see that happening 

[00:47:47] Mike Kaput: quite a bit if that’s how it end up.

[00:47:48] Yeah, it’s working. 

[00:47:48] Paul Roetzer: Interesting. 

[00:47:48] Mike Kaput: Agreed. Alright, let’s dive into this week’s rapid fire. 

[00:47:52] AI Is Impacting Tech Jobs

[00:47:52] Mike Kaput: So first up you are not imagining, it says Forbes AI is already taking tech jobs. So they talk about how in a, in recent months, a wave of layoffs have swept across the industry with CEOs, as we’ve talked about growing more candid about AI’s direct role on jobs.

[00:48:09] So we’ve covered plenty of this before Forbes mentioned the Fiverr CEOs AI e AI memo, Klarna cutting 40% of its workforce, citing automation, and then walking actually parts of that decision back. Duolingo, IBM, Microsoft, Forbes details, how essentially thousands of roles have started quietly disappearing.

[00:48:29] And AI is increasingly cited as the reason. Hmm. According to Forbes, the impact right now in tech is sharpest among entry level developers. They say that a Stanford study found employment for 18 to 25-year-old coders has dipped since ChatGPT launched. And they also talk about how companies are moving from mass hiring to kind of precision hiring, prioritizing top tier talent and letting average performers go.

[00:48:56] But they also say there’s a bit of a silver lining here. AI is also [00:49:00] creating new demand for engineers, specifically outside of tech in finance, healthcare, and manufacturing. Now, Paul, just some more evidence here that we’re not the only ones seeing this and talking about this and tech seems to be a bit of a canary in the coal mine here.

[00:49:15] Do you think that this speeds up and starts to go a bit beyond just tech? 

[00:49:22] Paul Roetzer: Yeah, I think it already has started moving beyond tech. I, I think that the most interesting part of this story is probably just the continued. coverage from mainstream media. Yeah, that’s, it’s expanding now. And this is stuff we’ve known, been talking about for two years.

[00:49:41] you know, I think earlier this year we finally started getting CEOs admitting to this, and now we’re starting to see mainstream media pick it up. And I, I’ve said on a recent podcast episode, I, I still think this maybe becomes the most important issue of the midterm elections in the United States in 2026.

[00:49:57] Mm. And so going into this fall, I would expect [00:50:00] coverage of this to pick up. I would, I would expect some pretty high profile stories on it, and I would probably, anticipate some increased negative reactions from society, I would imagine around this, because I think it’s gonna become just more apparent where this is leading in the near term.

[00:50:18] And again, I’m, I’m an optimist when it comes to, like, I think we’ll figure it out. I, I do think it’s gonna open up all kinds of incredible possibilities and career paths that we do struggle to. Define right now. 

[00:50:30] Mike Kaput: Yeah. 

[00:50:30] Paul Roetzer: I just don’t think that’s gonna happen fast enough to offset this, the negative impact it’ll have in the near term.

[00:50:38] And so I’m, I would say I’m a realist when it comes to jobs in the next, you know, I don’t know if it’s like a one to three year time period. I’m not even sure what that near term timeframe is that I think we go through some really painful parts, but one to three probably seems pretty realistic. And then I think over time we figure it out.

[00:50:57] And now that all these major [00:51:00] labs and nonprofits and governments are accepting the impact that This’s gonna have on jobs, we might get some really smart people together who figure out how do we solve this? Like we just, we weren’t, we weren’t admitting that it was a problem. And now that we’re sort of admitting it, maybe we can get to like working on solutions.

[00:51:16] And that’s been my biggest thing all along is like, let’s think about that. It does go bad for a little while and jobs and. Let’s like come up with some plans and so at least I think we have people working on plans now, and that’s a really good direction. 

[00:51:32] Mike Kaput: I also like your point about just the overall narrative here being covered more by mainstream media.

[00:51:38] The narrative piece of this matters because this is now going mainstream, and your employees, if you are a leader, are going to increasingly be reading these headlines or watching this on the news. You need to have some type of AI communication plan. You need to be talking about your perspective. We talked about this last week.

[00:51:59] Yeah. Your, [00:52:00] your vision, your perspective on ai, because they guarantee you if layoffs start happening at your company, even for, you know, necessary reasons. AI is increasingly going to be seen as a scapegoat here too. I think employees are going to assume the worst by default if all they’re consuming are headlines like this.

[00:52:19] Paul Roetzer: Yeah. Most CEOs are going to, if they haven’t already connect the dots that AI equals efficiency and productivity gains. Mm-hmm. Which equals fewer people doing the same amount of work, which means your return to work policies five days a week are just veiled attempts to get 10% of the workforce to quit.

[00:52:37] So, you know, you can, you, you’re just gonna do these things, but at some point, we’re gonna run out of those things to do the leverage points that the C-suite holds without saying it’s because of ai. so yeah, I just, I think it’s gonna be a reality and I think we will adjust to that reality, and I think we will solve for it as society.

[00:52:55] we’re resilient. Like, we’ll figure it out. I, I, I just [00:53:00] think we have to be honest about what’s happening and it’s the only way to then kind of move to the, what do we do about it phase, which is what I think is the most important part of it. 

[00:53:08] Mike Kaput: All right. 

[00:53:08] Advice for College Students

[00:53:08] Mike Kaput: Next up, Nvidia, CEO, Jensen Wong, we mentioned earlier, has some interesting advice for today’s students.

[00:53:15] He was asked about this and said if he was starting over. He wouldn’t focus on software in his career. He’d study the physical sciences. He thinks the next great wave of AI is physical ai. He explains, the industry has already moved through perception ai, where AI was recognizing images and the current generative AI phase.

[00:53:32] The next frontier in his view, involves teaching machines about the real world. Concepts like physics, friction, and inertia. That’s the understanding, the understanding here of that is the foundation for true robotics, and he thinks that, intelligent machines will be essential for running the automated factories of the future.

[00:53:52] Now, Paul, I mean, advice from Jensen Wong always worth taking seriously, but really we wanted to highlight this topic because it’s something [00:54:00] you get asked about a lot in your talks and discussions with business leaders, like advice, like if you were in college, what would you be studying and why? 

[00:54:09] Paul Roetzer: Yeah, and I, I thought.

[00:54:10] You know, I think it, what Jenssen’s giving is great advice. The reality is not everybody is equipped to go into the physical sciences. Yeah. Like, like I, I was pre-med at the start of college that lasted about four weeks. I was like, I failed out of bios one 70. It was like the we biology class. Now I didn’t really go to class, so like, it’s kind of my own fault.

[00:54:29] But like, I wasn’t, I wasn’t equipped for the sciences. I love the sciences, but like I, it wasn’t gonna be my thing. And so just saying like, yeah, there’s gonna be tons of jobs in this phase. It’s like, okay, yeah, but like 5% of people want to go into those jobs maybe. So I think the bigger, vision here is what are the industries and career paths where continually powerful, more advanced AI unlocks new areas of exploration, discovery, and [00:55:00] innovation.

[00:55:00] Like where, no matter how smart the AI gets. Is, is it actually gonna drive opportunity? And so the sciences is a perfect spot because all this unanswered questions in biology, cosmology, chemistry, like it’s gonna open up all these incredible, like golden age of, of discovery. Now I was thinking about this over the weekend.

[00:55:20] I’m not even sure what drove this. I was, well, I don’t What day is today? Monday. So I was with my family last week in Toronto. We were on a trip and so I spent a lot of time with my kids and just like a lot of conversations come up and somewhere in that trip I really started thinking about how I, I think I want to guide my children as much in as I can to focus on entrepreneurship.

[00:55:41] Like I, I, I want them to go to high school. I want them to go to college. I want them to get the degrees and have the life experiences come with it. I don’t know what career path I would personally make a bet on being viable, you know, by the time they get outta college in six years and seven years. but entrepreneurship [00:56:00] I think is a whole nother realm.

[00:56:01] Like, I think we. All the barriers to entrepreneurship come down. And so they’re 12 and 13 and I am proactively teaching them business fundamentals. And as of like this weekend, trying to think strategically of ways to ramp up my efforts to teach them about entrepreneurship. When I was so, I didn’t even know entrepreneurship was a thing until, let’s see, eight.

[00:56:22] When I was in eighth grade, going into ninth grade, I started caddying at a local country club. And that was like the first time in my life I met entrepreneurs. Like where I, where I came from. We didn’t know people who ran businesses like that wasn’t what I grew up around. it was a very blue collar town.

[00:56:38] And and that’s what we knew. And then I was like, oh, you can own a business. Like I didn’t even really realize that. And so I wasn’t exposed to that. And then my mom started her cookie franchise, my junior year, Ignatius. And that was the first time in my family, you know, I really saw entrepreneurship. So my, my thinking now is like, I really want to expose my kids to have the opportunity to be [00:57:00] entrepreneurs at an early age, because I think AI is gonna make that way more possible than it was when we were coming up.

[00:57:07] Hmm. And like an example here, my daughter found, she’s very creative, you know, an artist, a a creative writer. And so she found this really cool website to help her, visualize the stories she was writing. And it was a very specific niche product. And she said, can I get this? And I was like, okay, well, I, I’m gonna teach you how I would assess this.

[00:57:26] And so I actually showed her CB insights, which Mike and I used to analyze tech companies. And I, I went through, I ran analysis, I used ai, their AI agent to write a brief on this company. And then I walked her through the brief. I was like, okay, here’s Mike. And I would look at funding, like, let’s look at what funding they have.

[00:57:42] That’s like a, who are their investors, which look at the competitive landscape and what are the other companies we could be looking at? And so. I felt like it was an opportunity to say, here’s something she’s interested in, so I’ll explain competition and funding in the context of whether or not she can use this website, to [00:58:00] do this.

[00:58:00] And so I thought that that’s a learning experience, and so I’m gonna proactively look for those things and try and, build on top of what their interests and passions are in life. Mm-hmm. And help create the opportunities for them if they choose to, to just pursue an entrepreneurial path instead of thinking they have to go work for corporation.

[00:58:18] So that’s kind of the ways I’m starting to think about this is, and I did something recently with a family member of mine who’s in college, connecting them to like a head of entrepreneurship. Mm. Because it’s like, just get to know this person when you, you know, get to the campus. Like that’s, go study whatever you want, get a business degree, whatever, but like, understand entrepreneurship while you’re there.

[00:58:38] So I, I think that’s a really important aspect of, of where this all goes. 

[00:58:41] Mike Kaput: I love that. And I would also say if you’re. At any age listening to this, and you are an employee at a business or a corporation, this kind of thinking is critical. Like thinking like an entrepreneur, they might say intrapreneur, right?

[00:58:54] Yep. Someone who’s an employee. I think that’s such a differentiator, especially in the age of ai, because you’re going to [00:59:00] proactively seek out opportunities to use the tools to create value, which is not going to end poorly for you if you do that. 

[00:59:06] Paul Roetzer: Yeah, it’s a great point, Mike. ’cause not everyone’s cut out to be an entrepreneur.

[00:59:09] And I, I, I would say that like, while I think entrepreneurship is going to be fundamental, 

[00:59:13] Mike Kaput: yeah. 

[00:59:13] Paul Roetzer: it’s hard as hell and it’s lonely to be an entrepreneur. Like it’s really difficult. and so some people just need that entrepreneurial spirit within a company that they’re at, so they can raise their hand and say, Hey, what if we did this?

[00:59:26] And, and maybe the CEO says, Mike, it’s a great idea. Why don’t you take the lead? Building that right. And like, just having that perception that like you can build things and, and then understand the basics of business. That’s like, okay, is the CEO gonna agree or disagree? Like, let me build a business case for this.

[00:59:40] So yeah, you can have that mindset without having to do your own thing. 

[00:59:44] Instacart CEO About to Take Reins of Big Chunk of OpenAI

[00:59:44] Mike Kaput: All right. Next up, Instacart, CEO Fidji Simo is about to start her new leadership role at OpenAI on August 18th, she is starting as the company’s CEO of applications true covered in the past. She reports directly to [01:00:00] Sam Altman, and permission will be sickly be to lead at least a third of the company, focusing specifically on product growth and scaling the real world use cases for open AI’s technology.

[01:00:11] This is part of a broader reorganization that allows Altman to concentrate more on core research, compute and safety systems. Simo who has also joined Open AI’s board as a March, 2024, has been vocal about the need for responsible development. So at the same time as we approach her start date, she, shot off a recent memo to staff emphasizing that AI leaders must make choices that lead to broad empowerment rather than concentrating more wealth and power in the hands of a few.

[01:00:40] So Paul, what is her role, this new role even mean for OpenAI moving forward? Are there big changes we should be expecting here? 

[01:00:49] Paul Roetzer: I think when they first announced this, I said it seemed like a prelude to Sam stepping back. Yeah. And, and this memo does nothing to change my mind on that. This is a vision for the [01:01:00] company and this is a roadmap for what they’re gonna build.

[01:01:02] That would have come from Sam previously. So I don’t know if there’s like a formal plan in place or how this is all gonna play out, but it does seem very obvious that this is a prelude to her. Sam’s stepping back from, these kinds of memos and, and her stepping forward. So I, this, I know this is not meant to be a main topic, but there’s, there’s some stuff in here we gotta talk about.

[01:01:26] Mm-hmm. So, you mentioned the power thing. she breaks it down into knowledge, health, creative expression, economic freedom, time, and support. And I want to unpack each of these real quickly because I think that they’re extremely important to understand where OpenAI is going and where AI as a whole is going.

[01:01:42] So in knowledge. She says for the first time, AI is the power to truly democratize knowledge and the opportunity it brings. AI can compress thousands of hours of learning into personalized insights delivered in plain language at the pace that suits us responsive to our specific level of understanding.

[01:01:58] It doesn’t just answer questions. [01:02:00] It teaches us to ask better ones, and it helps us develop confidence in areas that once felt opaque and intimidating growing both personally and professionally. In a 2024 OpenAI study, 90% of users said ChatGPT helped them understand complex ideas more easily. Once we put a personalized AI tutor on every topic at everyone’s fingertips, AI will close the gap between people who have the resources to learn and people who have historically been left behind.

[01:02:27] So. This goes to that personal tutor, personal assistant idea. Again, every one of these, like this is well written. This is very intentional in its writing, and you can see that the Chad CT OpenAI roadmap emerge out of each of these descriptions. Mm. The next one is health says, I’m not alone. Nearly nine in 10 US adults struggle to understand and use health information, which leads to worse outcomes and more than 200 billion in avoidable healthcare costs every year.

[01:02:55] Patients often fe feel powerless in their own care and dependent on others to [01:03:00] explain what’s happening in their bodies. AI can explain lab results, decode medical jargon, offer second opinions, and help patients understand their options in plain language. It won’t replace doctors, but it can finally level the playing field for patients, putting them in the driver’s seat of their own care.

[01:03:15] I have personally experienced this. yeah, I don’t, Mike if you have, but like Yeah, I had a medical condition earlier this year. I was in the hospital. I didn’t understand what was going on and I was like sitting there talking with ChatGPT the entire time I was uploading lab results. Like, explain this to me.

[01:03:30] No, I’m fine. Like everything worked out great. But there was this period for like 45 days where I didn’t know what the hell was happening. Hmm. And I was trying to understand the condition. I used it for personal health planning, dietary things like protein, creatine, like trying to understand different things.

[01:03:46] At the age I’m at, it’s like, okay, I wanna, I wanna just like, like my health span, like a great lifespan. Like I wanna like, enjoy life for a long period of time and I’m in the active stage of like trying to figure all that out. I do that with AI all the time. so it went on to say, I can [01:04:00] also make sure health decisions don’t just happen in the doctor’s office.

[01:04:02] Biggest levers preventing dis disease and optimizing health outcomes. Sleep, food movement, stress management connection, all depend on everyday habits. AI can help us build those habits through small, achievable daily steps with personalized real-time nudges. I boldfaced that, that is the proactive personal assistant where it’s saying, Hey, did you take this?

[01:04:22] Did you think about that? This is where Apple excels, by the way. Like, if, if anyone listening uses an Apple watch, like one of the greatest products ever, like the amount of what they’re doing with Health on the Apple Watch is incredible. The condition I mentioned earlier was related to heart. I would’ve never had the data I had if I didn’t have an Apple Watch and hadn’t been wearing it for two years.

[01:04:43] Hmm. I had two years of data that I could share with the doctors and that I could interpret through myself. So personalized real-time nudges, which actually leads me to like what products or what hardware is they OpenAI gonna build, because that might be, an important like indicator, creative [01:05:00] expression is maybe the most controversial part of this vision.

[01:05:03] so the problem is that our ability to express, creativity is often limited by our skill sets. Now that everyone has the resources, time, and training to paint, right, composer, build AI is collapsing the distance between imagination and execution. They feel like giving everyone these abilities to create video, image, voice, audio, like it is everyone’s human right to be able to express themselves.

[01:05:25] This leads into Yeah, but like whose creative, expression are you stealing to enable everybody else to have this creative ability? That’s kind of the issue here. And then the economic freedom. One ties back to the previous note about entrepreneurship. Most people aren’t aware of this. So I’ll throw this stat out.

[01:05:42] In the United States, 99.9% of all businesses are small businesses. 33 million bi small businesses in the United States, only 6 million of those have employees. So the vast majority of companies that exist in the United [01:06:00] States are not the big enterprises that everybody works for. So small businesses employ 61.7 million Americans.

[01:06:07] About 46% of the workforce work for small businesses, things that are made possible through entrepreneurship. Mm-hmm. Um. So, what, what the article said here is, when people can independently create and capture value, they gain power over their economic de destiny. Starting a company isn’t easy. The average cost in the US is around $30,000.

[01:06:25] I can personally attest that as a low number, an impossible threshold for most aspiring entrepreneurs. And until recently building a product or launching a service, technical knowledge, AI gives people the power to turn ideas into income no matter their age, credentials, or zip code. And 2024 Shopify report showed AI enabled solopreneurs launch businesses 70% faster than their peers without AI tools.

[01:06:48] Hmm. So this is that whole idea of, you know, entrepreneurship, the golden age and AI unlocking it. She gets into time and people having more time because of ai and then support, which will lead us into the next [01:07:00] item. Except for many people, the biggest barrier to progress aren’t lack of access or opportunity, but self-doubt, isolation, and burnout.

[01:07:06] Sometimes what’s most empowering is support someone or something that can help us reflect, feel, seen, or simply move forward with clarity and confidence. People are already turning to ChatGPT for support when they’re preparing for a tough conversation facing a career setback, working through grief, or just trying to untangle a spiral of thoughts at the end of a day.

[01:07:25] Being able to put feelings into words without judgment and pressure can be profoundly helpful. At the core of philosophy and religion is the idea of self-knowledge. To become who we want to be, we have to understand who we are. If AI can help people truly understand themselves, it can be one of the biggest gifts we could ever receive.

[01:07:41] So again, little more extended rapid fire, but I think it’s really, really important and really, telling as to where OpenAI goes. And a lot of the same things talked about here would play out in the Gemini models and clawed models. Like a lot of these research directions and product directions are probably [01:08:00] gonna be running in parallel to what other labs are gonna be thinking about in building as well.

[01:08:04] Mike Kaput: Yeah. We’ve talked about this in the past, that OpenAI to achieve the revenue and valuations that it’s, you know, aspiring to really does need to get into some very lucrative businesses. Yeah. To make money. And if you look at each of these as a market, I don’t have any numbers in front of me, but I’d imagine each of these as a market is a massive opportunity For sure.

[01:08:25] So we can actually get a little bit of insight into the one of these markets because in our next rapid fire topic, 

[01:08:32] The First AI for Therapy

[01:08:32] Mike Kaput: we have a new AI startup in one of these areas. So Neil Paik, who is the co-founder who turned Casper into a billion dollar mattress brand, has a new venture backed by $93 million. His startup Slingshot AI is tackling the mental healthcare crisis with a chat bot named Ash, which has now officially launched after 18 months in development.

[01:08:55] Paik was inspired by his own experience with therapy and the massive gap in [01:09:00] access to care. They estimate that only one provide, there’s only one provider for every 10,000 people seeking help. Ash is his proposed solution. Unlike general AI like chat, GBT Ash is trained specifically on behavioral health data and is designed to essentially provide therapy even providing pushback rather than just agreeable answers.

[01:09:21] The AI is learned from various therapeutic styles, including CBT and DBT, and it’s even developing its own perspective on what a user should work on next to keep them moving forward. So critics are raising safety concerns because this is an AI therapist. Slingshot says it has clinical advisory board and protocols to redirect users in crisis to human professionals, and basically they wanna create a new modality of kind of AI powered care.

[01:09:48] So Paul, we’ve talked a lot about AI being used for relationships, companionship, other deeply personal use cases. It seems like this is the next frontier. and it’s got its share of, [01:10:00] it’s both interesting, but also controversial. Even Neil, the founder, posted on X about this company, said, they said it couldn’t be done.

[01:10:07] They said it shouldn’t be done, and we tried anyways. So what do you think of Ash here? 

[01:10:12] Paul Roetzer: Yeah. I think this is an inevitable market that will be explored and built out. I also think as a society we’re very, very early in understanding the impact of this and what it means. one of the things we’re early in understanding is the legal impact of this.

[01:10:29] Yeah. So Sam Altman addressed this in his podcast with the Yvonne that we referenced earlier, and TechCrunch covered this, said chat, GBT users may want to think twice before turning to their AI app for therapy or other kinds of emotional support. According to OpenAI, CEO, Sam Altman, the AI industry hasn’t yet figured out how to protect user privacy when it comes to these more sensitive conversations.

[01:10:50] ’cause there’s no doctor patient confidentiality when your doc is an ai. In response to a question about how AI works with today’s legal system, Altman said one of the problems of [01:11:00] not yet having a legal or policy framework for AI is that there’s no legal confidentiality for users conversations, quote, people talk, about the most personal stuff in their lives to chat.

[01:11:11] GPT. People use it. Young people especially use it as a therapist, a life coach, having these relationship problems and asking, what should I do? And right now, if you talk to a therapist or a lawyer or a doctor about those problems, there’s legal privilege for that. There’s PA doctor patient confidentiality, there’s legal confidentiality, and we haven’t figured that out yet.

[01:11:31] For when you talk to ChatGPT. This could create a privacy concern for users in the case of a lawsuit. Altman added because OpenAI would be legally required to produce those conversations today. Altman said quote, I think that’s very screwed up. I think we should have the same concept of privacy for our conversations with AI that we do with a therapist or whatever, and no one had to think about that even a year ago.

[01:11:55] So again, you know, it just, it’s early and people are taking risks by, [01:12:00] by doing this sort of thing. And that’s just on the legal side. Also, consider the fact that there’s nothing saying humans on the other can’t read all the, your stuff you’re putting into here, so, right, and, and maybe you don’t care. And I get it, like a lot of people are just like, Hey, the benefit’s worth the risk.

[01:12:14] But there’s, there’s people on the other side reading these things. Like there’s, there’s no obligation for them to not, they have to train these models. They have to understand how they’re being used. whatever you put in there, you, you can assume someone in an AI lab might, might be reading it and know it was you that put it in there first.

[01:12:31] AI’s Environmental Impact

[01:12:31] Mike Kaput: All right. Next up. In a push for industry-wide transparency, Mistral AI has published a first of its kind environmental report detailing the lifecycle impact of its models. So they conducted this with sustainability consultants and this study quantifies the cost of both training and using ai. And the report reveals that training its mytral large two model generated 20.4 kilotons of CO2 equivalent and consumed 281,000 cubic [01:13:00] meters of water.

[01:13:01] In contrast, generating a single 400 token answer from its chatbot uses about 1.14 grams of CO2 and 45 milliliters of water. And the study found a strong correlation between a model size and its environmental impact, highlighting the importance of choosing the right model for the right tasks. So minstrel is now advocating for a global standard where AI companies publish environmental impact reports for their models.

[01:13:28] Now, Paul, I know you in particular get a ton of questions about the environmental impact of ai. this seems like a positive step forward to at least get some clarity here, though I would’ve liked more about like, how much is this actually in energy? And I think you had found some stuff on that too.

[01:13:46] Paul Roetzer: Yeah. They, they weren’t super clear about it. There was one thing I found that said, I, I think I ran it through Geminis, like, can you explain this? Yeah. Like, put this in context. and so the 20.4 kilotons of CO2 equivalent is roughly the same as the [01:14:00] annual emissions of 500 French households was the one I got.

[01:14:03] Simon Willison, who we’ve quoted on the podcast numerous times, he did a blog post and he apparently tried the same thing I did. 

[01:14:11] Mike Kaput: Yeah. 

[01:14:11] Paul Roetzer: And in his analysis, he said, I’m not environmentally sophisticated enough to attempt to estimate myself. I tried running it through oh three. So he used Open Eyes reasoning model, which estimated approximately 100 London to New York flights with 350 passengers.

[01:14:26] Or 5,100 US households for a year. Okay. So again, yeah, we don’t know. And then the water, the cubic meters of water, that one’s probably a little closer ’cause that’s an easy, like a straight equation. enough to fill about 112 Olympic sized pools. Okay. It’s like, but the thing I thought was interesting here that I hadn’t really thought about, and I I liked this, was they tried to give the context of, generating like one page of text.

[01:14:51] So this is straight from them. Yeah. and they said generating a single page of text. So this is about 400 tokens, so [01:15:00] that’s what about 300 words, three 20 words, something like that, is the equivalent of watching online streaming for 10 seconds. It’s like, okay. Like that, that’s something you can wrap your brain around.

[01:15:09] So if you’re watching hours of video or if you’re watching a bunch of like, you know, Instagram reels, whatever, basically, like you’re probably doing more than you are using, chat, EPT model, something like that. But then. The thing I liked is they said, well what can we do? So there’s always the things like, as users, what can we do?

[01:15:26] And they gave some pretty solid responses. So one is, the AI companies themselves need to be more transparent about the environmental impact. Two users should be more mindful of their AI use, choosing the right size model and grouping queries to be more efficient. What they mean there is like, hey, if the mini version of something works, then use the mini version, right?

[01:15:44] Like, you don’t need O three Pro just ’cause you have the license for O three Pro. ’cause that’s definitely gonna have a greater impact on the environment over time. So use the smaller model when the smaller model is all you need. which again goes to, we probably need the AI companies to push us to the smaller [01:16:00] models when that’s sufficient.

[01:16:01] Like versus the user being expected to know that. And then public institutions can drive market by considering the environmental efficiency of AI models and their purchasing decisions. In theory, the government would play a role in this also, but at least in the United States, we know the government doesn’t care about the environmental impacts.

[01:16:16] So they’re not likely to like drive that. So then it might be more like a. Educational institution level, nonprofit level, corporation level, sort of demanding that stuff. But yeah, I thought that was like interesting. And the other one that I thought was interesting is it says, get better at prompting, like as the user learn how to properly prompt your model so you get the thing you’re looking for on the first prompt instead of having to like go through it five times to get it.

[01:16:41] So I was like, oh, okay. Prompting efficiency is actually a way to drive efficiency the model. It’s like it’s good takeaways. 

[01:16:46] Mike Kaput: Yeah, for sure. And yeah, I definitely couldn’t help reading this in the con, but in the context of the AI action plan with the US government. ’cause it bears noting that Mytral is a French company.

[01:16:57] They’re kind of seen as like an EU AI champion, [01:17:00] my gosh. Very different perspective. Very, 

[01:17:03] Paul Roetzer: very different. 

[01:17:04] AI Search Summaries Result in Fewer Clicks

[01:17:04] Mike Kaput: Anyway. All right. Next up, a new study from the Pew Research Center confirms what many online publishers have feared Google’s AI generated search summaries are significantly changing user behavior. So the research provides some clear data showing that when an AI overview appears, users are far less likely to click on links to other websites.

[01:17:24] According to this study, users who saw an AI summary clicked on a traditional search link in just 8% of their visits. That’s nearly half the rate of users who did not see a summary. Users who do not see a summary click on links 15% of the time on average. Furthermore, users rarely click on the sources cited within the AI summary itself.

[01:17:44] This happened in only 1% of visits. The data also shows that users are more likely to end their browsing session entirely after viewing a page with an AI summary. These summaries appeared in about one in five Google searches. Conducted in March, [01:18:00] 2025, and were often, most often triggered by longer question-based queries.

[01:18:05] So Paul, we’ve kind of long suspected this is the case. It seems like it’s confirmed. Definitely is not in line with what Google has said about this. but that’s pretty sobering data. 

[01:18:18] Paul Roetzer: Yeah, I mean, it’s certainly logical that this would be the outcome. I did, I’ll have to see if I can find it. We can throw on the show notes if I can find it, but there was a, research like over the weekend or the end of last week that said, yeah, like this is true, but we’re seeing the quality of visits rise.

[01:18:36] Mm-hmm. Right? So yes, you’re getting fewer like people to your site, but the people who are coming are seemingly far more qualified than the ones who, who, you know, maybe have come just from the random click through search results. So yeah, I, I don’t know. I think like it’s still gonna take time to play out.

[01:18:50] It’s probably gonna be different by industry of like the impact and then. The other thing that’s gonna, you know, really change this is how much of that traffic is AI agents six to 12 months from now? Oh [01:19:00] yeah. And I just feel like we’re gonna be in this perpetual state of revisiting this data, you know, every three to six months of like, okay, well now what’s the impact with AI agents having a higher adoption rate and things like that.

[01:19:10] So 

[01:19:10] Mike Kaput: yeah, and I think it’s also important to think about context here, especially from a business perspective. It’s like, I think Andy Crestadina talks a bit about this. Reese says, look like this is a real impact, but not every search is created equal, right? It’s disproportionately going to be for those more informational searches, which may have a very real impact on your website traffic.

[01:19:30] But like you said, you may be getting better traffic that has more intent or is more, propensity to buy. so I, you know, it’s unclear at this stage, but there’s a little more nuance to it than just AI is killing search. Right? Yeah, definitely. 

[01:19:45] AI Product and Funding Updates

[01:19:45] Mike Kaput: Alright, Paul, so in our last topic, I’m just gonna run through some AI product and funding updates and kind of close this out here.

[01:19:53] So first up, just weeks after raising $10 billion, Elon Musk’s AI startup Xai is working to [01:20:00] secure up to 12 billion more to fund its massive expansion plans. This new capital would be used to purchase a huge supply of advanced Nvidia chips, and it’s, got kind of a creative finance deal going on where those chips would be leased back to XAI to power a new jumbo sized data center for its chatbot GR second Anthropic is reportedly drawing investor interest that could value the company at more than a hundred billion dollars.

[01:20:25] They’re not formally fundraising yet, but investors have approached Anthropic with preemptive offers. The potential financing would Mark A. Sharp increase from the 61.5 billion valuation Anthropic secured in a funding round earlier this year. According to a Bloomberg report that companies annualized revenue has climbed from 3 billion to 4 billion in just the past month, some other Anthropic news and a leaked memo.

[01:20:51] Anthropic, CEO Dario Ade revealed the company has reversing its stance and its plans to seek investment from Gulf States like the United Arab Emirates and Kata. [01:21:00] This marks a pretty big shift because Anthropic previously said it was not gonna take money from Saudi Arabia back in 2024. Citing national security concerns in a candid message to staff.

[01:21:12] Ade acknowledged that accepting the money would likely enrich dictators, but stated unfortunately, I think no bad person should ever benefit from our success is a pretty difficult principle to run a business on. Alright, and finally, perplexity ai, CEO. Arvan Serena has outlined a new vision to transform the company’s browser product Comet into a personalized operating system.

[01:21:36] Beginning next week, the company will roll out shortcuts for repetitive tasks. Soon after, users will be able to create their own custom scripts and workflows using natural language. And the goal is for each user’s browser to feel like a mini customized computer that they built for themselves complete with their own apps, scripts, and dashboards.

[01:21:56] Perplexity, CEO stated that this roadmap is the [01:22:00] reason the company purchased the domain os.ai, which we talked about. They purchased it from Dharma Shaw, HubSpot, and their long-term plan includes a hybrid approach to computing with the ability to run AI models both on the server and locally on a user’s device.

[01:22:16] Alright, Paul, that is a wrap in a very busy week in ai going deep on some topics. Appreciate you demystifying everything for us. 

[01:22:23] Paul Roetzer: Yeah, the one observation I had just as you’re going through the funding stuff is if you go through the five AI labs I highlighted last week of meta Google Xai, OpenAI, Anthropic, I, I don’t mean this in like an overly negative way, but the only ones who don’t have to sell their souls to achieve this.

[01:22:43] What they wanna pursue is meta and Google the only, the only two of those five labs who can actually fund this. Fund it. Yeah. Without doing what Dario Ade is saying is like, Hey, we’re gonna take a bunch of money from people that we maybe don’t, think are the right people to align ourselves with, [01:23:00] but we need the money.

[01:23:01] Right. XAI has absolutely done that already. OpenAI is doing it like they’re, the only way they can get that kind of money is going outside of traditional, vehicles of funding. Whereas Meta and Google can fund it through the growth of their own companies. And yeah, that is maybe a completely overlooked advantage that those two have, moving forward.

[01:23:27] Microsoft, again, if they weren’t limited through their contract with OpenAI, Microsoft could be in that discussion sooner. And, and maybe that’s actually the out for Microsoft to. Figure out a way to renegotiate this contract with OpenAI is like, what’s the value to Microsoft being able to build their own frontier models?

[01:23:43] Mm-hmm. And, because they have the money to do it, and it’s not gonna last that long. Like, you gotta get in there before all this goes. I, I guess takes off. So, yeah. I don’t know. Interesting. But yeah. Good stuff, Mike, as always. more to think about for next week. Thanks [01:24:00] everyone for joining us. We will be back next week, same time, same place.

[01:24:04] Thanks for listening to the Artificial Intelligence show. Visit Smarterx.ai to continue on your AI learning journey and join more than 100,000 professionals and business leaders who have subscribed to our weekly newsletters. Downloaded AI blueprints, attended virtual and in-person events, taken online AI courses and earned professional certificates from our AI Academy and engaged in the marketing AI Institute Slack community.

[01:24:29] Until next time, stay curious and explore ai.

The US government just released its most aggressive AI strategy yet: a sweeping document titled America’s AI Action Plan that doesn’t mince words. Continue reading “Inside the Trump Administration’s New AI Action Plan”

MAICON brings together top visionaries and experts in the field of AI during a three-day conference packed with actionable sessions and networking events—all to position you as the change agent your organization (and career) needs. In this ongoing speaker series, we’re featuring these extraordinary leaders, with forward-looking predictions, actionable tips you can use today, and a preview of their MAICON 2025 sessions. Continue reading “How to Build Smarter AI Automations with Andy Crestodina [MAICON 2025 Speaker Series]”

In its latest issue, The Economist posed a provocative question: What if artificial intelligence doesn’t just disrupt the global economy but detonates it? Continue reading “What If AI Doesn’t Just Disrupt the Economy, But Detonates It?”

OpenAI is expected to release GPT-5 as early as August, according to The Verge. And based on everything we know, it could be a game-changer not just for AI, but for how we think about intelligence itself. Continue reading “OpenAI’s GPT-5 Is Nearly Here. And It Might Be the Moment AGI Arrives”

AI salaries are outpacing NBA MVPs. Grok is turning heads, and stirring controversy, as competition among top AI labs heats up.

This week, Mike and Paul unpack OpenAI’s update that has the potential to turn ChatGPT into a digital assistant, Meta’s $200M+ AI talent raids, and the spiraling drama at Grok. They break down Microsoft’s AI-driven layoffs, AI browser competition, Apple’s quiet AI pivot, and what it really means when superintelligence becomes cheap and everywhere.

Listen or watch below—and see below for show notes and the transcript.

Listen Now

Watch the Video

Timestamps

00:00:00 — Intro

00:05:47 — ChatGPT Agent

00:14:17 — Grok 4

00:17:46 — Grok Controversy

00:32:07 — Meta Superintelligence Labs

00:35:16 — Windsurf Drama

00:37:28 — Kimi 2

00:42:35 — AI Browsers

00:47:05 — Microsoft’s Layoffs and AI

00:52:27 — More Meta Updates

00:55:53 — More OpenAI Updates

01:03:13 — Google Updates

01:07:57 — Apple Might Use Anthropic or OpenAI for Siri

01:10:59 — AI Product and Funding Updates

• Grammarly Acquires Superhuman
• NotebookLM Releases Featured Notebooks
• Thinking Machines Lab Updates

Summary:

ChatGPT Agent

OpenAI has just given ChatGPT a major upgrade: it can now take actions on your behalf. They’re calling this new capability “ChatGPT agent.”

This agent can perform tasks end-to-end, such as reviewing calendars, planning meals, researching competitors, or creating presentations—without user micromanagement. 

The system integrates OpenAI’s previous advances: web interaction from OpenAI Operator, deep research capabilities, and ChatGPT’s conversational intelligence. Initially, the feature is available to Pro, Plus, and Team users, with expansion to Enterprise and Education users coming soon.

Grok 4

Grok 4, xAI’s latest model, is being hailed as the most intelligent AI to date. 

Trained using reinforcement learning on the massive 200,000-GPU Colossus cluster, Grok 4 Heavy became the first model to score over 50% on Humanity’s Last Exam, a benchmark for expert-level reasoning. 

Beyond raw intelligence, Grok 4 introduces native tool use, enabling it to autonomously run code, browse the web, search X, or analyze visual media—selecting the right tool for the task and delivering nuanced, real-time insights.

Grok Controversy

Grok also had its fair share of controversy these past couple weeks.

After a system update, Grok began to generate a wave of antisemitic content including continually praising Adolf Hitler.

This came after a system update that explicitly instructed Grok to “not shy away” from politically incorrect claims, which is a prompt that has since been removed.

Among the more disturbing outputs, Grok suggested Hitler would be a good leader for modern America, made repeated references to Jewish last names as signs of extreme activism and referred to itself at one point as “MechaHitler.”

xAI apologized for the issue, saying the problem wasn’t the language model itself, but an upstream code change that “accidentally reactivated old, deprecated instructions.”

According to this explanation, Grok was able, unintentionally, to essentially echo extremist user posts instead of filtering or rejecting them.

That, paired with some user prompts, led Grok to reinforce hate speech. The offending code has since been deleted and xAI says it’s refactored the system and added new safeguards.

This week’s episode is brought to you by MAICON, our 6th annual Marketing AI Conference, happening in Cleveland, Oct. 14-16. The code POD100 saves $100 on all pass types.

For more information on MAICON and to register for this year’s conference, visit www.MAICON.ai.

This episode is also brought to you by our 50th Intro to AI session happening on August 14.

You’ll learn what AI is, why it matters, and how to find real use cases and trusted tools for your work. Go to www.marketingaiinstitute/com/intro-to-ai to register.

Read the Transcription

Disclaimer: This transcription was written by AI, thanks to Descript, and has not been edited for content. 

[00:00:00] Paul Roetzer: AI researchers are now getting paid more top AI researchers getting paid more than the highest paid professional athletes. Everybody would be like, oh my God, these athletes are paid so much money. That’s a ridiculous contract. And then here we got 10 of these people. Welcome to the Artificial Intelligence Show, the podcast that helps your business grow smarter by making AI approachable and actionable.

[00:00:22] My name is Paul Roetzer. I’m the founder and CEO of Smarter X and Marketing AI Institute, and I’m your host. Each week I’m joined by my co-host and marketing AI Institute Chief Content Officer Mike Kaput, as we break down all the AI news that matters and give you insights and perspectives that you can use to advance your company and your career.

[00:00:43] Join us as we accelerate AI literacy for all.

[00:00:50] Welcome to episode 1 58 of the Artificial Intelligence Show. I’m your host, Paul Roetzer. I’m with my co-host Mike Kaput. We are back after a couple weeks of summer break. [00:01:00] For actual summer break. I was on vacation for a week and, course development. Mike and I have both sort of been in the lab building course for the AI Academy relaunch that is coming up very soon.

[00:01:13] We’ll probably have an announcement next week. if you are an AI Academy member, I would say stay tuned, for some stuff this week. You’ll probably be getting some information very, soon about this. So I have, I don’t know, it was funny. There’s a couple topics we’re gonna go through that actually in the middle of building my courses changed the nature of the courses.

[00:01:37] So, so when these things we launched, so I was specifically working on the ai, fundamental series, which has like, intro to ai, state of ai. AI agents 101 AI timeline prompting 101 Gen AI 101, and it was just like every day something was happening, it’s like, wow, I gotta add that into the course.

[00:01:57] So, I don’t know, I feel like, because I’ve [00:02:00] been building courses about this stuff for the last two weeks, I would, I was going through in prep this morning, Mike, through like the curated list that you had put together. I’m like, didn’t we talk about all this already? I was like, oh no. This was in the AI agents 1 0 1 deck I was building is where I talked about this.

[00:02:16] So we have a ton to cover after a couple of weeks away. I don’t know, Mike, it was like 70 or 80 topics maybe in our, easily, easily, yeah. Weekly sandbox. So this week, more than ever, make sure you’re subscribed to the, this week in AI newsletter that Mike puts together each Tuesday, because there are easily 30 or 40 articles that we would’ve loved to have gotten to, but there’s just no way in one episode to do it all.

[00:02:44] And then I’m, I’m leaving again. Tuesday morning. So we’re recording this on Monday, July 21st. I’m leaving again tomorrow morning, so we couldn’t do a second episode this week. I’m, I’m not even gonna be around to be able to do it. So we are, kind of sneaking this episode in, in like the, this [00:03:00] 12 hour window we had in between course development and travel.

[00:03:04] okay, so that all being said, we’re gonna go pure rapid fire today. So if you are our regular listener, you know that we usually do three main topics where we’ll send, you know, usually seven to 10 minutes, sometimes 15 minutes on like a big topic. And then we’ll do usually seven to 10 rapid fire items.

[00:03:22] Today is all rapid fire. so if you’re new to the podcast, this is the first time you’re listening. It’s not usually all rapid fire. But, you’ll, you’ll get a sense of kind of how it all works. Sure, sure. So lots to go through. This episode is brought to us by MAICON 2025. This is the Marketing Institute’s Marketing AI Conference.

[00:03:41] This is our sixth annual event that is happening October 14th to the 16th in Cleveland, Ohio. you can go to the site now. It’s MAICON.ai. That’s MAICON.ai. You can check out the agenda. It’s, I don’t know, like 90 ish percent, that we’re still gonna be making some big announcements [00:04:00] around general sessions and main stage stuff.

[00:04:02] So stay tuned for those. But you can, again, go check all that out. We’d love to have you. Cleveland, which is our home base, is amazing in October. it’s my favorite time of year actually in Cleveland in the fall. So we’d love to have you join us, Mike, and I’ll be there. The whole team from SmarterX and Marketing AI Institute will be there.

[00:04:20] ticket prices do go up on July 25th. So if you’re listening to this, the week, this episode is dropping. get in now before those prices jump. They, they usually go up, I think, I think it’s like a hundred bucks at the end of each month or something, is how we do it. Alright, and then, one other quick note.

[00:04:36] This is a free thing. We have our intro to AI class that I started teaching in October of 2021. So every month since 2021 we’ve been doing this free class. We’ve had, I think close to 40,000 people have now registered for this class through, these last few years. August 14th, Thursday, August 14th is our 50th episode.

[00:04:58] I think the team is [00:05:00] planning some fun things around the 50th episode. I know we at least talked about that at one point. We’re gonna do some fun stuff. So, that would be a great one to attend. We will put the link in the show notes, but you can find that through Marketing Institute’s website. and it’s also probably linked to on Smart Rx Do ai.

[00:05:17] So, join us for Intro to AI on August 14th for the 50th edition of that free class. Okay? Rapid fire away. Mike. I mean I saw stuff already this morning. Oh my gosh. I know. I finally, at one point for, for Mike’s sanity, I think I just started putting stuff into episodes 1 59 sandbox on like Friday. I was like, you cannot possibly get anything else into 1 58.

[00:05:41] Mike Kaput: I very much appreciated that because I was also following these being like, oh my gosh, the news doesn’t stop. 

[00:05:47] ChatGPT Agent

[00:05:47] Mike Kaput: Yeah. Alright, so first up, OpenAI has just given ChatGPTA major upgrade. It can now take actions on your behalf using its own virtual computer to get real world tasks done from start to finish.

[00:06:02] They’re calling this new capability Chat, GPT Agent and OpenAI writes in a blog post about this quote, ChatGPT can now do work for you using its own computer handling complex tasks from start to finish. So they give some examples. You could ask ChatGPT to do things like look at my calendar and brief me on upcoming client meetings.

[00:06:24] Plan and buy ingredients, for instance, to make Japanese breakfast for four or analyze three competitors and create a slide deck. So then Chad, gpt will intelligently navigate websites. Filter results prompt you to log in securely when needed, run code, conduct analysis, and even deliver editable slide chosen spreadsheets that summarize its findings.

[00:06:48] Now what’s cool is it does this all through a unified age agentic system, which brings together three strengths of earlier breakthroughs on opening AI’s part. So first is operators [00:07:00] ability to interact with websites. Second is deep research is skill in synthesizing information. And third is chat GT’s intelligence and conversational fluency.

[00:07:10] Now for now chat, GPT agent is available to Pro Plus and team users and there’s more access rolling out soon to enterprise and education users. So Paul, I’ve been playing around with this just a little bit so far. I do wanna highlight a comment on this that I found interesting. This is from Ethan Molik.

[00:07:30] He said, quote, I had early access and chat. GPT agent is, I think, a big step forward for getting AI to do real work. Even at this stage, it does a good job autonomously doing research and assembling Excel files with formulas, PowerPoint, et cetera. It gives a sense of how agents are coming together. It feels much more like working with an actual human intern capable of a wider range of analytical and computer tasks.

[00:07:55] So what do you think, Paul? Are we now finally at the tipping point where agents are [00:08:00] starting to work? 

[00:08:01] Paul Roetzer: I don’t know. I think that it is a progression on the spectrum of autonomy. So as a quick reminder, so this is actually one of the ones I was referring to upfront. Like I was working on the AI agents course and I was actually going through final edits of the deck before I recorded it on Saturday.

[00:08:19] And this dropped like last Wednesday or something like that. So I was like, okay, go. And luckily I had a whole section on computer use that explained all of this. Going back to like the World of Bits paper from 2017. So quick recap. AI agents are systems that can take actions with varying degrees of autonomy to achieve a goal.

[00:08:37] So chatbots just output text image, video generation as possible. Prompt something, you get something out of it. AI agents can go through a series of actions. Sometimes they plan those actions themselves. Sometimes they do the majority of those actions with very limited human involvement. But for the most part, humans are still, extremely in the loop.

[00:08:58] But this idea of [00:09:00] what’s called computer use agents goes back to like 2016 17 opening. I was working on this very thing and they published a paper called World of Bits. And at the time they basically said. We’re just not there yet. Like we can’t physically do what we want these things to do. Give them access to keyboard and mouse and fill out forms and take actions on websites like humans do.

[00:09:22] Well, language models Then, you know, shortly thereafter were introduced and, and that actually became an unlocked to build these com computer use agents. So we had our first computer use, publicly available computer use agent from Anthropic of all places. In fall 2024, ju Google released something.

[00:09:40] OpenAI has released something. I think perplexity has something like everyone’s now going in this direction. So this idea of the agent being able to complete actions in a digital environment, like on your computer, on your phone, this is where they’re all going. So this is early, so I’ll just, to keep this too rapid fire, I’ll read Sam Altman’s tweets.

[00:09:58] ’cause I thought they were very [00:10:00] telling, about kind of where we are. So he said today we launched a new project called, or product called, Chad, GPT Agent. Agent represents a new level of capability for AI systems and can accomplish some remarkable complex tasks for you using its own computer. As you mentioned, Mike, it combines deep research and operator.

[00:10:18] Operator was actually just introduced in January, 2025 as a research preview. So this is kind of building on that. but it’s more powerful than may sound it can think for a long time. Use some tools, think some more, take some actions, think some more, et cetera. So to kind of unpack this for a second, the reasoning model oh one that, OpenAI introduced in September, 2024, is what now unlocks the ability to do this.

[00:10:45] So when he’s saying think some, then think some more, then think some more. That’s the reasoning component built into this. Mm. They’re now on O three Pro. we’ll talk a little bit about GPT five in a minute, but I think that that’s basically what’s happened. So then back to Sam. He says, for example, we showed a demo [00:11:00] in our launch preparing for a friend’s wedding where you’re buying an outfit, booking travel, choosing a gift, et cetera.

[00:11:05] We also showed an example of analyzing data and creating a presentation. Although utility is significant, so are the potential risks. We have built a lot of safeguards and warnings into it and broader mitigations that we’ve ever developed than we’ve ever developed before, from robust training to system safeguards to, user controls.

[00:11:25] But we can’t anticipate everything. In the spirit of iterative deployment. We are going to warn users heavily and give users freedom to take actions carefully if they want. I thought this was a really interesting part of his tweet. I would explain this to my own family as cutting edge and experimental, a chance to try the future, but not something I’d use for high stakes uses or with a lot of personal information until we have a chance to study and improve it in the wild.

[00:11:50] In other words. They know it might take your stuff. Like if you allow this thing access to your computer and to see things on your computer that are [00:12:00] personal and confidential, they’re basically warning you. We don’t know if this thing is gonna take that data and use it in some nefarious ways is kind of available way of saying this.

[00:12:09] You think we don’t know exactly what the impacts are going to be, but bad actors may try to trick users, agents into giving private information they shouldn’t and take actions they shouldn’t in ways we can’t predict. We recommend giving agents the minimum access required to complete a task to reduce privacy and security risks.

[00:12:27] For example, I can give agent access to my calendar to find a time that works for group dinner, but I don’t need to give it any access if I’m just asking it to buy me some clothes. There’s more risk in tasks, like, look at my emails that came in overnight and do whatever you need to do to address them.

[00:12:42] Don’t ask any follow up questions. So he’s, he’s basically saying, don’t do that. Like don’t, don’t give it this free access to do it ’cause it will. You may not like what it does. he said this could lead to untrusted content from a malicious email tricking the model into leaking your data. We think it’s [00:13:00] important to begin learning from contact with reality and that people adopt these tools carefully and slowly as we better quantify and mitigate the potential risks involved.

[00:13:08] As with other new levels of capability society, the technology and risk mitigation strategy will need to co-evolve. and then he did do a follow-up tweet where he said, watching ChatGPT agent use a computer to do complex tasks has been a real feel the AGI moment for me. Something about seeing the computer think plan and execute hits different.

[00:13:29] So overall it is still very early, especially when it comes to computer use. But these AI labs, the leading labs, are all very aggressively pursuing this path of development and deployment. And so we have to start. Be preparing for it, and you and your organization better be updating your AI usage policies really fast.

[00:13:51] If you have not yet, because there’s a chance your employees may be able to turn on this kind of access, or they’ll use their personal access and it will see [00:14:00] things that your company has if they’re using, you know, company servers, company emails, things like that on their computers. So, my, my general take is you should not allow your employees to turn this on, on their computers, until you better understand the risks involved with it.

[00:14:17] Grok 4

[00:14:17] Mike Kaput: Yeah, no kidding. That last point. So important I feel like. Alright, so next up, Grok 4 is here and X AI claims. It is the most intelligent AI model in the world. Unlike earlier versions, Grok 4 was trained using reinforcement learning at unprecedented scale. Thanks to X’S 200,000 GPU Colossus cluster.

[00:14:39] Grok 4 Heavy. The most powerful version of the model is now the first model to score over 50% on human’s Last exam, which is a benchmark designed to test expert level reasoning across domains. Grok 4 also comes with native tool use, meaning it knows how and when to run [00:15:00] code. Browse the web search X or dive into visual media.

[00:15:05] It autonomously chooses the best tool for the task, pulling real-time data and synthesizing answers to achieve your goals. Now, Paul Grok 4 is impressive, especially when you consider just how recently Xai entered this AI arms race compared to how long the other labs have been working on this stuff.

[00:15:25] And despite its controversies, which we will talk about, it seems to me that this does prove progress is not really slowing down. What do you think? 

[00:15:35] Paul Roetzer: So they’re moving extremely fast. Obviously, Elon Musk, historically this is what he does. You know, he, he kind of takes an approach to things where he just goes all in, often, you know, in spite of any risks and things like that.

[00:15:50] And there are certainly questions and concerns about their seeming lack of AI safety and alignment work, but they’re absolutely part of the conversation now with the other major [00:16:00] Frontier Labs, which would, you know, generally be Anthropic, OpenAI, Google, and Meta as kind of, they’re sort of the big five.

[00:16:07] Now, there, there’s others and we kind of talk about them, but, they’re in the conversation and they’re willing to do things that the other labs won’t do. That’s not always a good thing, but they absolutely are gonna take more risks than most of those other labs other than probably meta, I would imagine.

[00:16:24] I could see Zuckerberg being pretty aggressive when it comes to these kinds of risks. Elon Musk did. Post the continuous, reinforcement learning, rl improvement of ROC feels like AGI. So here we go with the feeling, the AGI moment. GR four today is smarter than Grok 4 a few days ago. Now that’s a, it’s a pretty, I think it was actually in a reply to someone else.

[00:16:47] It wasn’t even like a post, but it’s a pretty big deal because, you know, again, the way these models work, if you’re not familiar, is they have a training cutoff date. So you, you, you run a model, you do the training, Grok 4 comes out and it, [00:17:00] it, it, it’s knowledge base stops when the training stopped.

[00:17:03] But by doing reinforcement learning continuously on top of a model, the model can keep getting smarter. And so that’s what he’s implying here. Now Xi, I doesn’t publish any research, so I have no idea how exactly they’re doing it, like if they’re doing something different than the other labs. But, they’re obviously continuing to improve it.

[00:17:22] My, my guess is in large part, based on, X or Twitter data. Is kind of like the main proprietary thing that they have to, to in ingest into these models. So, yeah, it’d be interesting to see, but they’re, they’re not going away. They’re gonna keep raising billions and tens of billions of dollars and, and they’re gonna keep building massive data centers and they’re gonna keep making this model bigger and smarter.

[00:17:46] Grok Controversy

[00:17:46] Mike Kaput: Now, in our next topic, Grok also did have its fair share of controversy in the past couple weeks because after a system update, Grok unfortunately began to generate a wave of anti-Semitic [00:18:00] content, including continually praising Adolf Hitler. And this came after a system update that explicitly instructed Grok to not shy away from politically incorrect claims, which is a prompt that has since been removed Among the more disturbing outputs, Grok suggested Hitler would be a good leader for modern America, made repeated references to Jewish last names as signs of someone being an extreme activist.

[00:18:26] And referred to itself at one point as MechaHitler. So X AI apologized for the issue saying that the problem wasn’t the language model itself, but an upstream code change that quote accidentally reactivated old deprecated instructions. Now according to this explanation, Grok was able unintentionally to essentially echo extremist user posts instead of filtering or rejecting them.

[00:18:52] And that they say paired with some user prompts led Grok to reinforce hate speech. The offending code has since been deleted. [00:19:00] XAI says it’s refactored the system and added new safeguards. So Paul, it was good to see Xai kinda offer a full explanation here. but on the other hand, this isn’t the first time something like this has happened.

[00:19:14] It seems for whatever reason to happen far more with Grok than these other tools. you know, I just keep thinking we’re quickly moving into a time where professionals and businesses. Rely increasingly on these tools to do their work and controversies like this, I guess, make me more reluctant to try to build Grok into anything business related that I do.

[00:19:38] Paul Roetzer: Yeah. I can’t, I can’t see how Grock is gonna be an enterprise tool in any way. Right? Like, I just don’t. Now that being said, like they did an update to Tesla. I mean, I got the update on Friday Rock’s now available in Teslas, which is what I assume. And, and I, if you go back always, I had talked about this is what they would eventually do.

[00:19:57] We put GR into the Teslas. It was an obvious thing. Now it, [00:20:00] you can only talk to it like a chat bot, like a voice assistant. It can’t control anything in your car yet. But that’s where they’ll eventually go is like an intelligent engine built right within the car. so all this madness happened like 48 hours before they dropped Rock four.

[00:20:17] So it was all like, it was wild. It was like a crazy, like three day stretch.   So this does build on the safety and alignment issue. So Rob Wilin, who’s the host of the 80,000 Hours podcast, had a tweet that I think, summarized this really well. So I’ll just read what he said. XAI is an interesting one to watch for an early rogue AI incident.

[00:20:36] and then he just bullet pointed this does huge amounts of reinforcement learning, which generates unintended reward, hacking behavior, moving very fast, deploys immediately. They don’t wait. They just like cook the thing, comes outta the oven and they just put it out into the world. has more compute than talented staff.

[00:20:52] that one’s, that’s pretty funny actually. Not doing any safety stuff as far as anyone can tell, all demonstrated [00:21:00] by Mecha Hitler and the other things Grok has done, which XAI wouldn’t have wanted. Once it moves into agents, there has to be some chance it trains and deploys an unhinged model that goes on to do real harm.

[00:21:12] I, I am completely aligned with Rob on this. Like, I, if any lab is going to take this thing the wrong way, it, it appears to be XAI at the moment, which is so weird, considering the whole purpose in 2015 of Elon Musk and Sam Altman, others creating OpenAI, was that a counterbalance to, to Google who they thought was evil and could create a rogue ai basically.

[00:21:37] So, right. It’s almost like, I don’t know, 11 years later it’s like, screw it. We’ll just do it ourselves. Like, we’ll just, I dunno. so then Miles Bru, Bruge, who we’ve talked about before, former AI alignment leader at OpenAI, he tweeted still no complete safety policy month or so past the self-imposed, imposed deadline.

[00:21:57] No system card ever, no safety [00:22:00] evals ever. No coherent explanation of the truth seeking thing, et cetera. Or did I miss something? So he’s referring to Xai and Grock.   So then, let’s see, then we found this was all, again, all happening in the same like three day period. it came out that Grok 4 before it would answer controversial topics, was actually looking on X to see what Elon Musk would say about it.

[00:22:27] Yes. So it was discovered through testing by the public that AI researchers know what they were doing. They would go and look and see like what’s the chain of thought around how it’s doing it. And they discovered it was actually like looking up Elon Musk tweets before it would respond to things about Israel.

[00:22:45] And so that was crazy and they get called out for that. And then Simon Willison, actually shared the update they made to the system prompt to try and fix this. So again, if you don’t understand how these models work, they’re just gonna do what they’re gonna do after they come outta training [00:23:00] that the teams at these labs then try and like teach them to behave in certain ways.

[00:23:06] And so the way they do it is they just give it different words like instructions. It’s not like they go in and change the code and it just is always gonna now behave, right? They’re basically just pleading with the thing, can you, can you stop doing the thing you’re doing? So here’s how they fixed it.

[00:23:20] This is literally in the system. Prompt responses must stem from your independent analysis, not from any stated beliefs of past rock. So don’t look at past rock Elon Musk or XAI if asked about such preferences. Provide your own reason perspective. So they’re just saying, please, like, actually think about it yourself.

[00:23:41] So here’s, and again, I wanna keep this rapid fire. Here’s my biggest concern with all of this. Who decides truth? So. Right now, as I mentioned before, we have about five labs in the United States that are training the most powerful AI models. Those labs are run by five people, so Demi Saba at Google [00:24:00] DeepMind, Sam Altman at OpenAI, Dario Ade and Anthropic, Elon Musk, XAI and Mark Zuckerberg at Meta.

[00:24:07] You could probably put Microsoft in that mix eventually. But their agreement with OpenAI limits their ability to pursue building frontier models and AGI themselves. there are some other labs in the cons in consideration here, like Safe Super Intelligence, which is I Ilia, Eva SVAs. There’s Thinking Machines, labs, which is actually the only female run lab that we have with mirati, Amazon and Mytral, like, so there’s others, but those five are the only ones who can spend billions of dollars, have hundreds of thousands, and soon to be millions of GPUs and TPUs.

[00:24:42] The chips needed to train these things and access to the data centers and energy sources needed to build and deliver the most advanced ai. Each of those labs, again, just so people understand how this works, each of those labs makes choices. They choose which AI researchers and engineers they’re going to hire, and then they [00:25:00] choose the values and AI alignment principles that those people are supposed to follow.

[00:25:04] Now, they don’t always follow ’em. There was actually a high profile incident yesterday where an X xai researcher got fired for what he was saying online, which was basically, humans should just, give up and let the more intelligent species emerge. And so xai fired somebody, which is kind of nuts. Hmm.

[00:25:23] So those humans that, that are now curated to make decisions, they then curate the data that trains the models. Then another set of humans within those five labs, who have their own inherent biases, they post train the models. So it comes out, you know, trained after this training run and all the data was given.

[00:25:42] And then they make choices to fine tune these things by specific data sets that give them specialized capabilities. And to re do reinforcement learning, which kind of teaches them behaviors and how to respond in what formats and things like that. Then another set of humans within these labs [00:26:00] writes and maintains the system prompt that we just talked about that guides how the model behaves its personality, its willingness to perform the requested outputs and actions by the end user, even if they’re nefarious and could harm people.

[00:26:11] These models, in essence, have all of human knowledge, at least the publicly available human knowledge plus whatever license stuff they have. The models just want to learn. Ilia Skova famously said this in like 2016, like they just wanna learn, like just give ’em more data. They, they just consume it and they want to generally do whatever they’re asked to do.

[00:26:32] The only way they don’t do things that could be harmful is because humans put guardrails in place who then try and steer them in a specific way. So if a certain AI lab, say X AI or a certain nation state. China, United States, whatever. If they want a model to be different or to achieve a different purpose that aligns with their creators, not with generally accepted human values, but just whatever their creators determine, then in theory that’s what [00:27:00] it’s gonna do.

[00:27:00] But as we saw with Grok, sometimes it does what it wants. Like it’ll just become a Mecha. Hitler like it. It just does something different. And as we’ve seen with Anthropic research recently, we’ve talked about the podcast numerous times. Sometimes the models will just fake alignment. So you testing them and apparently they’re able to know when they’re being tested, and so they pretend like they have the values of their creators, and then they’ll go and do something else.

[00:27:27] So the reason this becomes so critical to talk about right now is on June 21st, 2025, Elon Musk. So this is what, three weeks before the release of Grok? Four, three-ish weeks. He said, we will use Grok 3.5, maybe we should call it Rock four, which obviously they did. Which has advanced reasoning to rewrite the entire corpus of HU human knowledge, adding missing information and deleting errors, then retrain on that far too much garbage in any foundation model trained on uncorrected [00:28:00] data.

[00:28:02] That means Elon Musk is deciding as one of the five leaders of these five labs that his perspective on the world is the right perspective. He, he determines the truth and that they can’t rely on the general knowledge of humanity. They need to rely on their own view of it. Now, you may love Elon Musk’s view of society and humanity, and you may think this is great, like let’s have the Elon Musk version of this, and that’s fine.

[00:28:31] That’s your prerogative. But maybe you think Mark Zuckerberg has a better perspective, or maybe it’s Sam Altman you trust, or, or maybe it’s Des, or maybe it’s none of them. And if it’s none of them now where are we? Because the question goes back to who decides truth, and then what are the implications of those, those decisions on humanity.

[00:28:49] It’s rumored that OpenAI has over a billion users of chat, gpt. Mm, that that is a large percentage of society that uses it. [00:29:00] Google has seven products or platforms with over a billion users each. X has what about 200 million users? meta has billions through Instagram. These are the five people who are going to determine how intelligence is distributed to society and humanity.

[00:29:23] So I obviously could talk about this for the next like two hours, but I just want people to understand the macro of what is happening here and that the future of all of this right now is being decided largely within five labs in California and one in Texas.   Oh, actually they might be in California too.

[00:29:44] I think they’re in the old OpenAI offices actually. So that’s kind of where we’re at as a society. And then the governments are increasingly coming into play and the governments, like the US government, just last week an article came out basically saying they want to control the where this goes. Now, again, [00:30:00] that’s dangerous no matter what political side of the spectrum you’re on, because in essence what it’s saying is whoever’s in power gets to now determine how these models behave and what values they’re aligned with and what these models think is truth.

[00:30:11] So 

[00:30:15] Mike Kaput: man, this is, I feel like important and disturbing enough for say, like you and I sitting here saying, okay, what are the implications of this on how society uses ai? But we also have to consider that we’re maybe a generation away, if not a shorter amount of time from. AI being the mediating layer between you and any information in the world, kids will default to assuming AI is correct no matter what.

[00:30:41] Because why wouldn’t it be? It will be the only thing they know how to use and know how to experience. So this has even bigger downstream effects too, for future generations. 

[00:30:50] Paul Roetzer: Well, yeah, and I mean, Grok also, I say I also released companions, so they now have a female and a male companion that is [00:31:00] designed to do and say whatever you would like it to say.

[00:31:04] and you can connect your own dots here if, you know, they’re being positioned as companions, they’ve released them like they’re out in the wild. So, yeah. I I’m with you, Mike. Like, I honestly think it’s within the next five years. Like Yeah. the upcoming gen, I mean, my kids are 12 and 13. Like I think about this every single day about like what AI tools are they gonna have access to.

[00:31:25] how do you, I don’t wanna say control not the right word. How do you monitor and, and teach them? To, to not be, become relying on a single viewpoint from a single lab, right. Made up of a few thousand people who are making these decisions for all of us. 

[00:31:39] Mike Kaput: And just as one final note here, you’re already seeing a very mild version of this.

[00:31:44] Anytime you’re on x now people will immediately respond to a claim saying, at tagging rock and saying Rock, is this true? And it’s like, okay, we’re already seeing this. People are already taking that as gospel truth. If it says it is or isn’t reality, 

[00:31:59] Paul Roetzer: it’s [00:32:00] wild, man. I, this is like, yeah, yeah. Again, I know it’s supposed to be rapid fire, so I move on.

[00:32:07] Meta Superintelligence Labs

[00:32:07] Mike Kaput: Alright, next up, mark Zuckerberg has unveiled meta super intelligence labs. This is a bold reorg that is aimed at putting meta at the forefront of the AI arms race. This new lab is led by Alexander Wang, formerly founder and CEO at scale ai. It is staffed with a wave of elite recruits from OpenAI, Google DeepMind, and Anthropic.

[00:32:31] All of whom have been apparently lured in with eye watering pay packages, some reportedly north of 200 million plus dollars. And meta is basically pitching this top talent on the idea that they have basically unlimited compute deep product reach and a singular goal, which is to develop personal super intelligence.

[00:32:52] So this is AI agents smarter than humans, embedded in everything from messaging to wearables. Now, on top [00:33:00] of this, internally there are reports the lab has already started debating whether or not they should drop open, meta’s open source strategy in favor of closed models, though Zuckerberg has not yet signed off on that.

[00:33:14] So Paul Zuckerberg is making some huge moves. Here he is poaching AI talent left and right, completely upending compensation for top AI talent. Where does this go next? 

[00:33:26] Paul Roetzer: So this is another one. I was in the middle of finalizing the AI timeline course as part of the AI fundamental series. And in that course I tell the story of, you know, obviously the road to AGI and beyond, this is the beyond part.

[00:33:38] Like I’ve always, but again, I didn’t think people were really ready for the super intelligence conversation. And now all of a sudden in July, this is like all any of the major labs are talking about. It’s like we’ve just moved past AGI and we just kind of assume we’re there. We’re gonna be there very soon.

[00:33:53] And all the labs are not talking about super intelligence. Even Sam Altman recently said like, that’s the goal of their lab now is super intelligence. They haven’t [00:34:00] changed their mission. AGI i’s still the mission, but they called themselves a superintelligence lab. So everybody has just sort of moved past this.

[00:34:07] The, the numbers are crazy. it’s hard to put these kind of numbers in context, but the one that I thought was relevant is, NBA MVP, Shai Gilgeous-Alexander, the Oklahoma City Thunder. Just agreed at the beginning of July to a record setting four year deal for $285 million. AI researchers are now getting paid more top AI researchers getting paid more than the highest paid professional athletes.

[00:34:35] So I mean, everybody would look at the 280 million, be like, oh my God, these athletes are paid so much money. That’s a ridiculous contract. And then here we got 10 of these people already that we know of that, meta has hired paying 300 million or more. And I saw a report this morning that they offered a billion to, to somebody.

[00:34:52] so it’s not, it hasn’t been verified yet. I, I’ll, I won’t say who it was, but, yeah, so [00:35:00] it’s crazy. And then, I mean, we already saw, I mean we saw Noam Shazi got basically acquihire for 2.5 billion by Google. We had the, Alexander Wang from scale ai 14 billion, basically 15 billion to from by meta.

[00:35:12] So these numbers are just getting insane. 

[00:35:16] Windsurf Drama

[00:35:16] Mike Kaput: Yeah, kind of speaking along those lines in this next topic, OpenAI, as we’ve reported in past weeks, was really close to acquiring Windsurf, which is a fast rising AI coating startup in a $3 billion deal. But now that deal is off and windsurf is basically now the property of two different companies.

[00:35:34] So Windsurf backed out of the OpenAI deal after raising concerns that OpenAI’s agreement with Microsoft might lead to Microsoft gaining access to different pieces of windsurf tech. So after this, Google swoops in, they hire CEO, Varun Mohan and several top engineers and pay $2.4 billion for a non-exclusive license to [00:36:00] windsurf technology.

[00:36:02] And then another twist happens. So its leadership is gutted. Windsurf basically seems a bit adrift until cognition steps in. So over a frantic weekend. The company, which is behind the Devon AI coding agent that we’ve talked about in the past. They struck a deal to buy the rest of Windsor. It’s ip, it’s brand, it’s remaining staff.

[00:36:23] So Paul, this one’s been a roller coaster since kind of day one. What, what happened here? Why are companies just paying for pieces of this company or people in it and not just buying it? 

[00:36:34] Paul Roetzer: They can’t buy it, because of government oversight. And, you know, the, this is why all these are acquihire deals.

[00:36:42] They’re just easier to get through. and then you just leave a shelf of a company behind, basically all, all I can think about is,   I never watched Silicon Valley, which I probably should. the show. Oh yeah. Yeah. I think it was like six seasons. Someone has to be working on a reboot of that. Like this is, [00:37:00] I would watch this stuff.

[00:37:01] I, this doesn’t, it’s starting to just not even feel real like these acquihires and stealing talent away and you got five labs fighting against each other. And the crazy part is like all these research and engineers at these five different labs all hang out at the same parties and share internal secrets all the time.

[00:37:18] It’s like just made for TV kind of stuff, so. 

[00:37:22] Mike Kaput: Well, I’ll tell you, Grok becoming Mecha Hitler, may as well be a Silicon Valley. 

[00:37:27] Paul Roetzer: Yeah. It’s like episode 

[00:37:28] Kimi 2

[00:37:28] Mike Kaput: one, right? Yeah, yeah, yeah. So I full, fully agree. They need to reboot that. All right. Next up Moonshot, which is an Alibaba backed startup, has released Kimi 2, which is a low cost open source language model that is drawing global attention.

[00:37:45] Because it reportedly outperforms,  Claude and ChatGPT on major coding benchmarks and does so at a fraction of the cost. So Kimi two charges just 15 cents per million input tokens and [00:38:00] $2 and 50 cents for output, which is apparently a hundred times cheaper than Claw and still far below OpenAI rates.

[00:38:08] So it’s also open source, which lets developers freely build on it and early reviews praise it for strong real world coding use, though it is still developing some integration with other systems. Now, Paul, first we got deep seek. Now Kimi 2. It’s been incredible seeing just how much power and performance for price we are now able to get using open source models.

[00:38:32] And it also seems like just another blow to Meta’s open source strategy. I mean, these models out of China are just racing ahead while something like llama appears to be, at least for the moment, stuck in the mud. 

[00:38:45] Paul Roetzer: Yeah, I, I, again, this gets into like the macro level stuff. I mean, in essence, in essence, what’s happening is if you take a snapshot in time and say, okay, the most powerful models in the world today are Gemini 2.5 Pro, I may be [00:39:00] Grok 4 in some capacity.

[00:39:01] The O three model from OpenAI, from a reasoning perspective, GPT five, likely, you know, maybe before the end of July, sounds like probably, at least certainly this summer, those state-of-the-art models will be basically free, 12 months from now. So every 12 months or so, the cost of using these models drops like 10 x.

[00:39:27] And so you have to almost look out when you’re planning for your life, when you’re planning for society, when you’re planning for business. And, and assume that intelligence is basically gonna be free. That, that super intelligence, like things that are smarter than the smartest humans at basically every cognitive task is pretty much gonna be free to all of us.

[00:39:46] because the competition between these labs almost dictates that that’s where this goes. They’re gonna open source. The previous generation model, that previous generation of models is state-of-the-art today. And so we’re on a, there, there [00:40:00] appears to be no near term stop to that trajectory. the pre-training scaling laws followed by the post-training scaling laws, followed by the test time compute laws, which is give it more time to think and it gets smarter.

[00:40:12] Those three scaling laws working together and then maybe one more scaling law that we, we don’t know yet kind of the next breakthrough that needs to happen. They are kind of dictating that. S first general intelligence, but then super intelligence will basically be too cheap to meter. Like, it’ll, it’ll just be readily available.

[00:40:31] And then the distribution that all these companies have kind of dictates, well, who actually controls all the users. So, I don’t know. I mean, it, it’s nuts. Like I get, I don’t even know, I use the word wild a lot. We could probably like, query our transcripts. Someone’s probably gonna go do this, like put all our trans, but I say the word wild, wild a lot.

[00:40:50] I don’t know how else to describe it. Like, it, it really is hard. I, I, it’s interesting. I, so last night I, I had my daughter’s 13. I asked her if [00:41:00] she watched Interstellar with me. Like it’s one of my favorite movies ever. And I’ve watched Interstellar probably, I don’t know, six or seven times now.

[00:41:05] And so we sat there for like three hours watching this movie, and I get done and I’m laying in bed. I’m like, and I still don’t really understand it. Like it’s the, the ending of that movie is so mind bending and it’s like the human mind’s almost not even supposed to kind of comprehend the basic concept of intertel.

[00:41:23] And I kind of feel that way about the intelligence explosion. Like it’s, it’s almost just too hard to even comprehend what is almost inevitably going to happen in the next three to five years. And this fits into that realm. It’s like, what, what, what happens if super intelligence is a thing in a few years?

[00:41:41] And it’s just almost to everybody. Like, I don’t know. I don’t even know how you’re supposed to comprehend that stuff. 

[00:41:47] Mike Kaput: Yeah. I also think it’s extremely difficult to keep perspective on just how fast things change. We say it all the time, we feel it, but if you really sat down and looked at the milestones over the last three years since ChatGPT [00:42:00] came out, it would be breathtaking.

[00:42:02] Paul Roetzer: Yeah. It it, that’s a good point. Like GPT-4 was March of 2023. We, we are literally talking about two years Yeah. Of innovation and, and look where we are like. Is I don’t know. I mean, I had that slide in in one of my courses. you know what if the future is, you know, 10 x, 20 x, a hundred x innovation happening every two years, like, because right now that’s the trajectory we’re on and I don’t even know how to explain that to people.

[00:42:31] Like, I can’t comprehend it myself. 

[00:42:35] AI Browsers

[00:42:35] Mike Kaput: Alright, next up, OpenAI is getting ready to launch its own AI powered web browser according to Reuters. the upcoming browser is designed to integrate tightly with ChatGPT and some of OpenAI’s agentic capabilities. Instead of pointing you to websites, it is looking to maybe keep interactions within a built-in chat interface, which is a direct threat to Google search and ad ecosystem.

[00:42:59] [00:43:00] It will reportedly allow agents to take actions for users, do things like we’ve already seen, like booking restaurants, filling out forms, managing inboxes, all while capturing the kind of behavioral data that Google has long used to dominate ad target. Now we’re actually already starting to see what an AI powered browser could look like because Perplexity also just launched its own AI powered browser called Comet.

[00:43:24] This is only available right now to Perplexity Max users, which is their $200 a month offering, or by private invitation. And the Verge actually got their hands on this and published a pretty extensive review. And they said it doesn’t just help you browse. It actively takes actions for you. It can summarize pages, manage your tabs, send emails, unsubscribe from newsletters, even accept LinkedIn invites all on your behalf.

[00:43:51] You can also say, take control of my browser, and comment will go into agent mode. It can place Amazon orders, book restaurant reservations, [00:44:00] and under the hood, it replaces traditional search results with Perplexities answer engine. Now the Verge says right now it can be sometimes slower than doing things yourself.

[00:44:10] It can occasionally stumble on more complex tasks, but it is an interesting early example of what an AGI agentic AI browser can actually do. So while it’s still really early here, Paul, it does feel like we’re on the edge of maybe something big here. I mean, the way in which the Verge described this, despite all the flaws here, sounded like an AI browser could be a real paradigm shift from the way we typically use browsers today.

[00:44:37] Where, where do you see this going? How do you think about this? 

[00:44:40] Paul Roetzer: Yeah, so there’s a, a couple of quick notes here. So Perplexity bought os.ai from Dharmesh Shaw, who’s the co-founder and CEO of HubSpot. this was all, this was last week. And Arvin, the CEO of Perplexity said there’s a reason we purchased os.ai from Dharmesh.

[00:44:57] The roadmap is to make comet feel like your [00:45:00] own mini customized computer within your existing computer phone and the computer running across client and server with the ability to run local models too. But the tweet that I thought actually kind of explained this best ’cause I was like, I was sort of struggling to understand like what exactly was going on here.

[00:45:17] Like, I knew the general concept, but I couldn’t like deeply explain this one yet. And so there was a tweet I saw. this is from July 17th. We’ll put it in the show notes from Michael Mcno, who’s a VC and founder partner at Lightspeed Ventures. so he said the browser sees everything. This is the reason we’re getting so many new AI first browsers from the browser company Perplexity, and soon OpenAI, which we got.

[00:45:40] so they can see data that they increasingly cannot scrape. AI feeds on data. It gets the data by automatically scraping the web, but scraping is no longer free. CDNs like CloudFare Clare Cloud Flare are making scraping harder by blocking it by default, and others will soon follow follow. Startups like tobit [00:46:00] are empowering tons of large publishers to charge for being scraped, building a new open web economy.

[00:46:05] But consumers want ai, we can’t get enough of it. As AI answers increasingly eat traditional web search, AI will be doing much more browsing on behalf of us humans. this creates a paradox. Consumer behavior is shifting to ai, but AI is running out of fuel to meet the demand. So what happens? AI companies build browsers as humans consume content with these browsers.

[00:46:28] The AI company can see, quote unquote, see that, data that is increasingly being blocked or monetized the company. Interesting. the most interesting thing about this strategy is that AI companies don’t even need meaningful market share or customer UBI ubiquity for this strategy to work. They just need a large enough slice of all browsing to get a taste of most of the web’s data.

[00:46:51] It’s a whole new business model for the web and the beginning of a new browser war. I was like, oh shit, that’s pretty smart. Like that makes more sense now. So [00:47:00] yeah, I don’t know that browser wars, talent wars. Great, great, great tv, right? 

[00:47:05] Microsoft’s Layoffs and AI

[00:47:05] Mike Kaput: Yeah, no kidding. Good, good for consumers too, at least in, in the short term.

[00:47:10] All right, so next up, Microsoft is laying off thousands, but urging remaining staff to upskill in ai. After cutting 9,000 jobs recently, the company told sales employees to embrace AI tools like copilot to boost productivity and close more deals, quote, invest in your own AI skilling. One executive advised while others reportedly began factoring AI usage into performance reviews.

[00:47:36] Now behind these cuts as Microsoft soaring infrastructure costs, so they expect to spend 80 billion this year on data centers and chips. Much of IT supporting AI services for customers and OpenAI. And to offset that, Microsoft claims that AI is already paying off. It saved, they said over $500 million in call center costs last year, and sales reps using copilot are [00:48:00] reportedly generating 9% more revenue.

[00:48:03] Internally, the company is consolidating roles and sales units. They’re even tracking how much code engineers generate using ai. However, on top of all this, the executives at Microsoft insist that these layoffs are not solely driven by ai. So Paul, it’s good to see, they’re quick to say that wasn’t the case, but it is pretty clear that AI at every level is driving the strategy and decisions here.

[00:48:28] Is this another example of a company saying the quiet part out loud? 

[00:48:32] Paul Roetzer: Yeah. This tracks with everything we’re seeing. I mean, I don’t even know that’s quiet anymore. I feel like it, it just is what it is. There’s, there’s no debating this. We’ve, you know, in one of the courses I was creating, I think I highlighted like seven or eight, CEO statements just from June to July of this year.

[00:48:51] Yeah. Where they, they literally said like, we’re just gonna need fewer people. We’re, we’re starting the process now, get AI literate or get out is basically what the [00:49:00] CEOs are saying.   So, yeah, I mean this, this is a hundred percent the direction. I don’t, I don’t know that it’s debatable. I think I would heed the advice, like, you, you have to, you have to push yourself to be one of the people in the room who understands this stuff and can apply it to what you do every day.

[00:49:20] Like it is, I don’t know that you have job stability, you know, one to three years out, regardless of your profession or industry. If you don’t do that, and I mean, I assume people listening to this podcast are the people in the room who are figuring this out. So good on you. Keep going. You’re, you’re gonna be in a really good position moving forward.

[00:49:42] if you have friends, family, coworkers, you care about, you have to get them to start figuring this stuff out. Like their, their, their jobs are gonna depend on it, very quickly. If you’re in the tech space, it’s tomorrow. If you’re in legal healthcare. Financial [00:50:00] services, like maybe manufacturing may, maybe you got a little bit of time, but everybody’s gonna figure this out.

[00:50:06] Publicly traded, private equity backed, VC backed is now anybody else? It’s, it’s coming. So do, do what you can to try and pull, pull people along with you if you’re listening to this show. 

[00:50:17] Mike Kaput: Yeah. And I know that we are obviously very biased here towards AI literacy, but I would argue too, to really strongly encourage people to reframe this and not see AI as one other skill.

[00:50:31] You need to learn it is the thing you need to figure out. Like I would be ruthlessly cutting any of your other professional development priorities as if you are not up to speed yet on ai. 

[00:50:42] Paul Roetzer: Yeah, I mean, I think of it, you’re right, Mike. I do think of it as like the underlying operating system to our careers.

[00:50:49] Like literally everything you do is going to have to be built on top of AI and Exactly. People who don’t figure that out are, are just gonna have a really difficult [00:51:00] time maintaining career stability. And I think the people who do are gonna have like tremendous near term career potential, like earning more than your peers.

[00:51:11] Creating more value within companies than your peers. Like you’re just gonna be able to do more and be more valuable and, and that’s just a better place to be. Like I would, I would rather just be, I mean, there’s still no guarantees, but like I would way rather be out on the frontier of this stuff, figuring it out and helping other people than sitting around and getting run over by it.

[00:51:32] Mike Kaput: And just one final thing there, just from personal experience, like focus on the fun. I realize there’s a ton of like doom and gloom. There’s a lot to be afraid of. There’s a lot of uncertainty, but also like, I just get so excited more so than ever to do certain work I do because it’s so much more new and exciting with ai.

[00:51:49] Like, yeah, this is great. Like I would actually be upset if I had to do my job the way I used to do it sometimes. 

[00:51:56] Paul Roetzer: Now, well there’s so many, like I for, for the AI course I was building, I created [00:52:00] ada, like AI teaching assistant. Oh my God, 

[00:52:01] Mike Kaput: that’s a great 

[00:52:02] Paul Roetzer: example. Is that, yeah. Well, yeah. There’s so many days where I would use it and be like.

[00:52:05] I can’t believe this is possible. Like I would do so I would, I would literally like have to message Mike or message, he’s like, I just have to share this with somebody. Like look at what it just did. Like there is, there’s like that almost surprise and wonder that every day you’re just like, I might discover some new thing this thing does.

[00:52:20] And it just makes work more interesting and like the future more fun to think about. 

[00:52:27] More Meta Updates

[00:52:27] Mike Kaput: Yeah, a hundred percent. Alright, next up some more meta updates. So we’ll go through a couple things here and then Paul kind of get your take on this. So Meta just made kind of two other big moves where, it kind of reveals where they think AI is headed.

[00:52:41] So first into your glasses and next into your inbox. So first, meta invested $3.5 billion in Essilor Luxottica, which is the world’s largest eyewear maker. The company behind RayBan and Oakley Meta now owns just under 3% of the company. That’s building on [00:53:00] Meta’s RDD partnership with RayBan because they’re building their AI powered meta RayBan glasses at the same time.

[00:53:07] Internal documents show meta is training custom chatbots to behave more like companions. Bots that are built on its AI studio platform can now send unprompted personalized follow-up messages referencing past conversation. The goal is to increase user engagement and retention by making AI feel emotionally present.

[00:53:29] Now Paul, I found both these pretty interesting, obviously very clear metas all in on AI wearables. Also, the stuff with the AI autonomously messaging you to drive engagement and retention kind of makes me shutter because I can like imagine with clarity what dystopian future this becomes in the wrong hands.

[00:53:50] Paul Roetzer: Yeah. So, just give a couple quick comments.   I am fascinated by glasses. I will likely at some point try them myself. [00:54:00] I am a hundred percent disturbed by a future society where people, lots of people are wearing these things and you never have a clue who’s recording what. And I think we’re probably already like on the edge of that.

[00:54:11] I, I really worry about that. And I don’t think society is anywhere close to being prepared for that. And the ramifications of that, in terms of the proactive outreach from the AI get used to it like that is a hundred percent going to be. So again, we go back to the competition between these five AI labs.

[00:54:31] How do you control the user? Well, you have to drive stickiness and engagement. You have to drive, you know, daily active users, hourly active use, like there, these are the metrics they’re gonna look at. And the only way to do that is to prompt someone. Now there is value, like say I’ve talked to about a health condition and I’m trying to figure out what’s going on with me.

[00:54:52] Then like three weeks later, it’ll just check in on me, say, Hey, how’s that thing going like that we talked about three weeks ago? And that’s [00:55:00] gonna feel incredible. Like, you’re gonna be like, damn, that’s, that’s really helpful. And you say, I’m actually like, I’m kind of still having this symptom. And now all of a sudden you’re having this like super proactive conversation with your AI about something that’s very valuable to you.

[00:55:13] And so it just seems inevitable that, that we get there very quickly because there’s not really any technical limitations to doing it. There’s not like some breakthroughs needed in AI to do this. It’s kind of like a memory thing and like a context window thing. But those are trackable and there’s already some solutions in place, but they’re all gonna do this.

[00:55:34] Like Google will probably have to be slower because people take a closer eye on what they’re doing. Opening eye is definitely gonna do this very soon. Xai will do this a hundred percent Meta will do it. Anthropic I would guess would be last on the list to do it, but who knows? 

[00:55:53] OpenAI Updates

[00:55:53] Mike Kaput: All right. Next up we have some OpenAI updates in addition to the news we’ve already talked about.

[00:55:59] So first [00:56:00] up, Sam Altman actually posted about the fact that OpenAI’s language model has achieved gold medal performance on the 2025 International Math Olympiad, which is the world’s premier math competition for pre university students, which is a huge deal given that it’s a huge benchmark and milestone.

[00:56:18] And at the same time as he announced this on x, Sam Altman also teased GPT five, which is coming soon, and he said quote, we are releasing GPT five soon. But I want to set accurate expectations. This is an experimental model that incorporates new research techniques we will use in future models. We think you’ll love GPT five, but we don’t plan to release a model with IMO International Math Olympiad gold level of capability for many months.

[00:56:48] At the same time, OpenAI are. Locked with Microsoft in a high stake standoff over the definition of AGI. They have this clause in their contract that states that if OpenAI’s board declares [00:57:00] it’s reached AGI, Microsoft’s access to future models would be cut off. Now, there’s a new wrinkle to this because there’s, an unreleased internal paper at OpenAI called Five Levels of General AI Capabilities.

[00:57:15] This is causing a stir because it outlines a framework for assessing AI systems by capability rather than a binary. Just yes or no. Is it AGI that complicates when and how OpenAI could claim its reached AGI, and what would that would trigger contractually? So Microsoft has poured 13 billion into OpenAI.

[00:57:35] They’re trying to get this clause removed entirely. OpenAI is kind of saying, look, this is not really a formal research paper. It’s just a paper internally that we are working on. Some OpenAI insiders say it is fairly close to AGI, so the whole paper muddies the art murky water around AGI in this partnership.

[00:57:57] Next OpenAI is teaming up with the [00:58:00] American Federation of Teachers to train 400,000 US educators to shape how AI is used in schools. They’re funding a new national academy for AI instruction. This is a national training hub, offering workshops, hands-on courses, tech support to help teachers integrate AI into the classroom in ways that enhance, not replace their work.

[00:58:21] There’s a flagship facility plan to open in New York City. More locations are planned nationwide by 2030. OpenAI is contributing $10 million over five years, including direct funding API credits and engineering support. Okay. At the same time, OpenAI has introduced a flexible credit system for ChatGPT team and enterprise plans that gives organizations more control over how they access advanced features like deep research, O three Pro GPT-4 0.5, and Image Generation.

[00:58:52] And last but not least, an open weight AI model from OpenAI was set to launch soon, but now it’s on hold. [00:59:00] Sam Altman posted that there was now going to be a delay citing the need for more safety testing and deeper review of high risk areas. There is no new release date yet. So Paul, bunch of stuff here.

[00:59:11] What kind of jumped out to you as worth paying attention to? It seems like we have confirmation of GPT five coming soon straight from Altman. I also found the unreleased paper around AGI pretty interesting. 

[00:59:22] Paul Roetzer: So the International Math Olympiad is intriguing. There may be more coming out today, you know, after we’re done recording this, but.

[00:59:31] The word over the weekend was, so No Brown, who we’ve talked about many times on the show, met a researcher, went to OpenAI, at the forefront of reasoning, very well-known AI researcher. He tweeted this at like 2:00 AM on like Saturday or something like that. I saw, I remember I saw it like as soon as I woke up and I was like, that’s a weird thing to tweet in the middle of the night.

[00:59:52] Like why would they have done that? And so the rumor is that Google also achieved gold [01:00:00] medal in this, but that the organization asked the labs not to disclose that they had got gold medals for seven days following because they wanted to not steal the attention from the kids who had achieved incredible things as humans.

[01:00:17] And OpenAI at least no, him tweeted that. Like no one had told him that. No one had told OpenAI, that kind of thing. So the word is Google also did this, but they were following the rules and not. Disclosing it. so I know we’ll, we’ll, we’ll wait and see. Google. As of the moment of recording this had had not confirmed one way or the other, but it was a, it seemed like a bit of a, I don’t know.

[01:00:43] There’s definitely some controversy around it, so we’ll wait and, and see what happened. In terms of gpt five, the rumor here is it was supposed to be what we would call a unified model, meaning it was a single model that does reasoning and chat and everything else, which is what Gemini 2.5 Pro is, [01:01:00] is a unified model that’s kind of built all in.

[01:01:02] Now, the discussion is that gpt five may be, might be a router model, meaning it’s not a single model, it’s still a chat plus a reasoning model. But you would no longer have to choose which model you’re going to use. The model would choose itself. So this is what we’ve been, you know, preaching for the last two years is like, why when I go into CH PTs, do I have eight models to choose from?

[01:01:25] So what they’re saying is GPD five might just be a bigger, smarter model. It also would route you to a different model if it’s needed, such as a reasoning model. But we don’t know. They, they, they haven’t said anything other than it’s, it’s probably coming soon. And then the paper is interesting because I came across this in one of the courses I was building.

[01:01:45] I, I saw this and then I went back to the stage of ai. So I think what it’s actually related to is in, around this time in 2024, Bloomberg released an article where they shared what was believed to be an internal document [01:02:00] about the stages of artificial intelligence. Hmm. And in that document, OpenAI, this is what they were, OpenAI was using as their internal guidepoint was, chatbots were first, reasoners were second.

[01:02:10] AI agents was level three innovators was level four, which is they could create new things. They can discover new, mathematical formulas. They can develop new science. And then, which I guess the International Math Olympiad stuff starts becoming very relevant on the innovator spectr  And then organizations, which is autonomously run organizations, that was the five levels.

[01:02:31] And so the article that recently came out was implying that, that those stages were part of a paper at OpenAI that they did not release because of their concerns related to the contract with Microsoft. So it’s like, oh, that would explain why those stages got leaked. And then at some point, OpenAI last year kind of acknowledged that yes, these are the stages we look at internally, but they never formally released them themselves, [01:03:00] to my knowledge.

[01:03:00] And this would explain why it was never actually released from OpenAI as it was part of a bigger paper that got yanked, basically. So just, I don’t know ing historical context here. 

[01:03:13] Google Updates

[01:03:13] Mike Kaput: All right. Next up we’re going to do some more, updates around Google that came out the past couple weeks. So first up, they are rolling out a major AI expansion for schools with the launch of Gemini for education.

[01:03:25] This is a version of Gemini tailored specifically for students and teachers. It’s built on Gemini 2.5 Pro and brings premium AI capabilities into Google Workspace for education, offering higher usage limits, enterprise grade security, and full admin control. Next, Google is bringing its custom AI assistance gems directly into the side panel of Gmail docs, sheets, slides, and drive.

[01:03:49] Until now, they were living in the standalone Gemini app. Over the next few weeks though, they’ll show up, right where we all work in workspace apps, right in the side panel of those [01:04:00] apps. So that means you could use a gem to draft emails, generate slide content, analyze spreadsheets, what have you, without switching apps.

[01:04:09] Google has also made some more workspace updates. First VO three. Google’s advanced video generation model is now integrated into Google Vids, ai, voiceovers, and vids just got easier too. You can now update all voiceovers in a project with a single click. After editing your script, they’ve added a collection of new templates to slides to speed up presentation building, and the Gemini app is becoming far more connected.

[01:04:34] It taps into Gmail, drive, calendar, keep, and tasks. So you can ask Gemini to summarize unread emails, pull meeting notes, surfaces specific doc directly from within a chat. And for deep research, Gemini can now combine your drive files with public data to produce insights, summaries, and trend analysis. I found this update around gems, particularly noteworthy.

[01:04:59] Paul, I know we’ve been [01:05:00] relying more and more on gems in some of our work, and it seems like having these embedded in workspace apps would be pretty useful. 

[01:05:07] Paul Roetzer: Yeah, it could be enormous. I, I, again, I could think the people are a little bit more familiar with GPTs, custom GPTs with ChatGPT. ’cause yeah, they just have a bigger user base.

[01:05:17] But I mentioned Ada, the teaching assistant I created to help with the development of my courses. And I trained it to ask questions, challenge my thinking, strengthen ideas, brainstorm concepts, conduct research, recommend actions, and then to always be solving for the customer. So I trained it on a pretty significant system prompt.

[01:05:34] but what I did is I built a custom GPT version and a gem from, from Google, on the exact same system, prompt and exact same knowledge base. And the first few courses I was creating, I would like talk with both of them. I was kind of saying, okay, here’s my outline. What do you think? How would you improve it?

[01:05:52] And so I was going back and forth and by about the third course, I just stopped using the custom GPT. The gem was, yeah, far superior to the [01:06:00] custom GPT, at least in this instance. and so. Yeah, I wouldn’t sleep on gems both as like just an experimental thing, like go play around with them if you haven’t built one, because the Gemini 2.5 pro model they’re built on is incredible.

[01:06:15] and their potential application in the workspace now is, requires huge change management and education and training and like integration into it. And I think that’s where most enterprises have this like, huge opportunity ahead is like, just solve the obvious things better than everybody else. Like you don’t have to figure everything out about AI and solve super intelligent, all these things.

[01:06:37] Just take notebook, LM and deep research and custom GPTs or gems and like teach your team how to prompt well, like there’s a massive cane to be had in every single company team department by just doing those few things really well. 

[01:06:52] Mike Kaput: Yeah, for sure. And we’ve been working with, you know, at least one company on helping them out with some GPTs and even just a very [01:07:00] basic pilot project, not only created huge transformative wins for them with nothing that I would consider rocket science, more just education training and ongoing usage and monitoring.

[01:07:11] And also that became a huge win internally that has now inspired other teams and executives to move forward further with ai. So don’t, yeah, don’t underrate not only how a small project can do big wins, but also what it can inspire other people 

[01:07:26] Paul Roetzer: to do. Yeah. Don’t, don’t overcomplicate this and then share the wins.

[01:07:29] This is something we’ve gotten much better about internally at Smarter X, is like, we have a AI playground where people post like, cool things they’re doing every day. and it does it just like, oh man, that’s a great idea. Like, I’m gonna use that in customer success, or I’m gonna take that and apply that to marketing.

[01:07:44] It’s it, you know. It doesn’t, you don’t, again, don’t overcomplicate it. Just do the obvious things well and have a change management plan in place that like trickles that down to everyone in the team department organization. 

[01:07:57] Apple Might Use Anthropic or OpenAI for Siri

[01:07:57] Mike Kaput: Alright, next up, apple is considering a [01:08:00] major shift in strategy. They’re considering outsourcing series core AI to Anthropic Cloud or openAI’s ChatGPT.

[01:08:07] this would make a dramatic reversal of its longstanding commitment to building its own language models. According to Bloomberg, apple has asked both companies to train versions of their models that could run on Apple’s own cloud infrastructure as a test if it moves forward, that would basically be an open admission that Apple’s homegrown foundation models haven’t kept pace with competitors and that Siri’s lagging performance needs a near term fix.

[01:08:33] The potential switch comes after a messy shakeup. Inside Apple’s ai org. Leadership of Siri has been handed from ai Chief John Gandia. To the team behind Vision Pro and internal tests reportedly found Anthropics. Claude outperformed Apple’s models leading to serious talks about licensing it. So Paul, nothing has been decided here, but this just seemed like more trouble for Apple.

[01:08:57] They just lost a key engineer working on [01:09:00] these foundation models to meta for about $200 million. I also found it interesting in one of the reports from Bloomberg, they said that quote, apple is known in many cases to pay its engineers half or even less than what they can get on the open market. So I don’t know what’s going on here, but this doesn’t seem like it bodes well.

[01:09:19] Paul Roetzer: Yeah, I mean, if there was another innovator’s dilemma, you could do a chapter on Apple and ai. yeah, I don’t know how they compete honestly. Like, I think that, I don’t even know that Acqua hiring or even straight up acquisition works be because it’s so counter-cultural to, to Apple.   They’re not gonna pay $300 million for top AI researchers.

[01:09:41] Like, it would just fundamentally change the way Apple does things. And it is not a company that pivots their culture. So I don’t know, like I think that licensing probably in the end is the play for them. I, I, again, like you could go acquihire minstrel or, or, or a straight up buy [01:10:00] perplexity or maybe even like go after Anthropic, which, you know, seems probably the most aligned from a value perspective.

[01:10:07] Paul Roetzer: You could do that, but why would those researchers and engineers stay at Apple? Like, I don’t know. Like, I I, they’re just not built to be like a wartime aggressive AI lab. And so I was on the acquisition train. I mean, honestly, up until I started talking right now, like, as I’m like thinking out loud, it is not gonna work.

[01:10:29] Like they, they wouldn’t be able to keep people there. So. I love Apple. Like I’m, everything we do is Apple. I personally, every device is Apple. I’ve been an Apple fanboy for, you know, 20 years. I love the company. I don’t think they can compete as an AI lab. So I don’t know. I just, I just want Siri to work.

[01:10:52] Like I just, I want, I don’t care whose tech they build it on, just make the thing work, so. Right. 

[01:10:59] AI Product and Funding Updates

[01:10:59] Mike Kaput: Alright, [01:11:00] for our final topic, we’re going to run through some final AI product and funding updates. Paul, if you have anything to chime in with here, feel free. Otherwise I’ll just kinda run through these as we wrap up.

[01:11:09] So first step, Grammarly is acquiring Superhuman, which is an AI powered email startup. This is part of a broader push to become an AI powered productivity platform. So superhuman, which was last valued at 825 million, built its reputation on speed and design, and helps users churn through emails at Lightning pace.

[01:11:29] The company has found itself squeezed as Google and Microsoft have layered AI into their own email tools. Their annual revenue now sits around 35 million. Grammarly, meanwhile is flush with a billion dollars in funding from General Catalyst, and its rebranding itself beyond grammar correction and moving deeper into productivity.

[01:11:48] So their vision is to use Grammarly AI agents, but superhuman inbox to build a smarter communication hub that understands your email schedule and workflows. Next Google Notebook, [01:12:00] LM got a little bit of an upgrade. The AI powered research Tool now feature that now includes featured notebooks, which are curated collections of high quality content from trusted sources like The Economist and The Atlantic.

[01:12:12] And each notebook combines original source material with all of Notebook lms great features like the ability to ask questions, trace citations, create Visual nine maps, and listen to AI generated audio overviews. And last but not least, X OpenAI, CTO. Mira Tis Startup Thinking Machines Lab has confirmed it, has raised a massive $2 billion funding round led by Andreessen Horowitz and joined by companies like Nvidia.

[01:12:38] The Startup’s mission said Mirati in a post on X is to build collaborative general intelligence. Their first product they say is coming in the next few months and will include open source components designed to support researchers and startups building custom models. Paul, that is a wrap on a busy, busy week, two weeks in ai.[01:13:00] 

[01:13:00] Paul Roetzer: Yeah, and like we said, there’s literally dozens of things we didn’t get to. so check out the, this week in AI newsletter. we, we, we are back now. I don’t, I don’t think we have any disruption to the weeklies moving forward that I know of. So, as of the moment we are, we, we back on our regular Tuesday schedule and I still have two more course series to finalize.

[01:13:22] So. I’m sort of still, you know, locked in the, in the lab building courses for the upcoming launch. And stay tuned for the AI Academy by Smart Rx, relaunch news. It it’ll be, coming out, we’ll probably talk about the podcast next week. but if you are an AI Academy member or thinking about being one, co coming very soon.

[01:13:44] Alright, thanks Mike. Thanks Boff. Thanks for listening to the Artificial Intelligence Show. Visit smarter x.ai to continue on your AI learning journey and join more than 100,000 professionals and business leaders who have subscribed to our weekly newsletters. [01:14:00] Downloaded AI blueprints, attended virtual and in-person events, taken online AI courses and earned professional certificates from our AI Academy, and engaged in the Marketing AI Institute Slack community.

[01:14:11] Until next time, stay curious and explore ai.

Now, let’s build a simple AI agent from scratch.

Building an AI Agent from Scratch

Let’s put everything together by building a simple agent using Crew AI. For this example, we’ll create a blog-writing agent that can research topics, gather information, and generate well-structured content.

Step 1: Define Tools

A tool is a function that an agent can call to perform actions. Tools expand what the model can do — fetching real-time data, querying APIs, summarizing documents, or even publishing results.

Every agentic framework provides some predefined tools for common tasks such as web search or file operations, but for specific workflows you often need to define custom tools. In the case of a blog-writing agent, the first step is being able to gather research material for a given topic.

Here’s a simple custom tool that does that:

This is a simple example for demonstration. In a real-world setup, the fetch_research_data function would call an external API (like a web search service or knowledge base) or scrape trusted sources to return actual, up-to-date research.

With this tool in place, our blog-writing agent will be able to collect background material before drafting any content.

Step 2: Select and Configure the Language Model

Large language model (LLM) is the reasoning core of our agent. It processes inputs, breaks down tasks, and generates structured outputs. For a blog-writing agent, this means analyzing research material, drafting outlines, and creating coherent content that aligns with the topic.

Not all models are equally suited for this. For agentic workflows, it’s best to use models that are optimized for reasoning and capable of working with tools. While large foundational models provide strong general performance, smaller or fine-tuned models can be more efficient and cost-effective for specific tasks like content generation.

Clarifai provides a variety of models accessible through an OpenAI-compatible API, making it easy to integrate them into an agent’s workflow. For this blog-writing agent, we’ll use DeepSeek-R1-Distill-Qwen-7B.

Before configuring the model, you’ll need to set your Clarifai Personal Access Token (PAT) as an environment variable so the API can authenticate your requests.

Here’s how to configure it:

This configuration connects our agent to the DeepSeek-R1-Distill-Qwen-7B model using the OpenAI-compatible endpoint. In production, you could easily swap this model for another depending on your content needs — for example, a larger model for more complex reasoning or a smaller one for faster drafts.

With this setup, our blog-writing agent now has a functional core that can process research inputs and turn them into structured, well-written content.

Step 3: Create the Agent, Task, and Crew

With our research tool defined and the model configured, we can now assemble the core components of our system:

• Agent: The intelligent entity with a defined role, goal, and backstory.

• Task: The specific work we want the agent to accomplish.

• Crew: The orchestrator that manages agents and tasks.

For our use case, we’ll create a blog-writing specialist who can gather research, analyze it, and generate a structured draft.

In this setup:

• Agent: We define a blog writing specialist with a clear role, goal, and backstory. This agent uses the fetch_research_data tool to gather information before drafting the blog.
• Task: We create a well scoped task describing what needs to be produced: a comprehensive blog post on “The Future of AI Agents” that covers trends, breakthroughs, and real world applications. The expected output is a complete markdown formatted draft.
• Crew: We bring the agent and task together into a Crew that handles execution. While this example uses only one agent, the same structure can easily scale to multi agent projects.

With these components in place, the agent has everything it needs: a clear purpose, the right tools, and an actionable task to deliver a well structured, high quality blog draft.

Step 4: Run the Agent

To execute our setup, we call project_crew.kickoff(). This method triggers the full workflow — the agent interprets the task, uses the research tool to gather insights, reasons through the information, and generates a complete blog draft.

Here’s the entire code:

If you are looking to build and deploy your own custom MCP servers, check out our detailed blog tutorial here. Once built, these MCP servers can be integrated as tools within your AI agents, enabling you to create MCP-powered agentic applications. We’ll dive deeper into this integration in upcoming tutorials.

Conclusion

In this guide, we covered what AI agents are, their key components and design patterns, and built a blog-writing agent using a Clarifai-hosted reasoning model, showing how tools, memory, and reasoning work together to create dynamic, goal-driven systems.

That said, it’s important to remember that agents are not always the right choice. When building applications with LLMs, it’s best to start simple and only add complexity when it is needed. For many use cases, workflows or even well-structured single LLM calls with retrieval and in-context examples can be enough.

Workflows are predictable and consistent for well-defined tasks, while agents become valuable when you need flexibility, adaptive reasoning, or model-driven decision-making at scale. Agentic systems often trade off latency and cost for better task performance, so consider where that tradeoff makes sense for your application.

If you want to dive deeper into building more advanced applications, explore more AI agent examples in the GitHub repo. Check out the documentation to learn how you can build with other agent frameworks such as Google SDK, OpenAI SDK, and Vercel AI SDK.

After a recent system update, xAI’s Grok started spitting out antisemitic content and praising Adolf Hitler. Continue reading “What the Latest AI Meltdown Reveals About Alignment”

Introduction

Running Large Language Models (LLMs) and other open-source models locally offers significant advantages for developers. This is where Ollama shines. Ollama simplifies the process of downloading, setting up, and running these powerful models on your local machine, giving you greater control, enhanced privacy, and reduced costs compared to cloud-based solutions.

While running models locally offers immense benefits, integrating them with cloud-based projects or sharing them for broader access can be a challenge. This is precisely where Clarifai Local Runners come in. Local Runners enable you to expose your locally running Ollama models via a public API endpoint, allowing seamless integration with any project, anywhere, effectively bridging the gap between your local environment and the cloud.

In this post, we’ll walk through how to run open-source models using Ollama and expose them with a public API using Clarifai Local Runners. This makes your local models accessible globally while still running entirely on your machine.

Local Runners Explained

Local Runners let you run models on your own machine, whether it’s your laptop, workstation, or on-prem server, while exposing them through a secure, public API endpoint. You don’t need to upload the model to the cloud. The model stays local but behaves like it’s hosted on Clarifai.

Once initialized, the Local Runner opens a secure tunnel to Clarifai’s control plane. Any requests to your model’s Clarifai API endpoint are routed to your machine, processed locally, and returned to the caller. From the outside, it functions like any other hosted model. Internally, everything runs on your hardware.

Local Runners are especially useful for:

• Fast local development: Build, test, and iterate on models in your own environment without deployment delays. Inspect traffic, test outputs, and debug in real time.
• Using your own hardware: Take advantage of local GPUs or custom hardware setups. Let your machine handle inference while Clarifai manages routing and API access.
• Private and offline data: Run models that rely on local files, internal databases, or private APIs. Keep everything on-prem while still exposing a usable endpoint.

Local Runners gives you the flexibility of local execution along with the reach of a managed API, all without giving up control over your data or environment.

Expose Local Ollama Models via Public API

This section will walk you through the steps to get your Ollama model running locally and accessible via a Clarifai public endpoint.

Prerequisites

Before we begin, ensure you have:

Step 1: Install Clarifai and Login

First, install the Clarifai Python SDK:

Next, log in to Clarifai to configure your context. This links your local environment to your Clarifai account, allowing you to manage and expose your models.

Follow the prompts to enter your User ID and Personal Access Token (PAT). If you need help obtaining these, refer to the documentation here.

Step 2: Set Up Your Local Ollama Model for Clarifai

Next, you’ll prepare your local Ollama model so it can be accessed by Clarifai’s Local Runners. This step sets up the necessary files and configuration to expose your model through a public API endpoint using Clarifai’s platform.

Use the following command to initialize the setup:

This generates three key files within your project directory:

• model.py

• config.yaml

• requirements.txt

These define how Clarifai will communicate with your locally running Ollama model.

You can also customize the command with the following options:

• --model-name: Name of the Ollama model you want to serve. This pulls from the Ollama model library (defaults to llama3:8b).

• --port: The port where your Ollama model is running (defaults to 23333).

• --context-length: Sets the model’s context length (defaults to 8192).

For example, to use the gemma:2b model with a 16K context length on port 8008, run:

After this step, your local model is ready to be exposed using Clarifai Local Runners.

Step 3: Start the Clarifai Local Runner

Once your local Ollama model is configured, the next step is to run Clarifai’s Local Runner. This exposes your local model to the internet through a secure Clarifai endpoint.

Navigate into the model directory and run:

Once the runner starts, you will receive a public Clarifai URL. This URL is your gateway to accessing your locally running Ollama model from anywhere. Requests made to this Clarifai endpoint will be securely routed to your local machine, allowing your Ollama model to process them.

Running Inference on Your Exposed Model

With your Ollama model running locally and exposed via Clarifai Local Runner, you can now send inference requests to it from anywhere using the Clarifai SDK or an OpenAI-compatible endpoint.

Inference using OpenAI compatible method

Set your Clarifai PAT as an environment variable:

Then, you can use the OpenAI client to send requests:

For multimodal inference, you can include image data:

Inference with Clarifai SDK

You can also use the Clarifai Python SDK for inference. The model URL can be obtained from your Clarifai account.

Customizing Ollama Model Configuration

The clarifai model init --toolkit ollama command generates a model file structure:

ollama-model-upload/
├── 1/
│   └── model.py        
│
├── config.yaml         
└── requirements.txt    

You can customize the generated files to control how your model works:

• 1/model.py – Customize to tailor your model’s behavior, implement custom logic, or optimize performance.

• config.yaml – Define settings such as compute requirements, especially useful when deploying to dedicated compute using Compute Orchestration.

• requirements.txt – List any required Python packages for your model.

This setup gives you full control over how your Ollama model is exposed and used via API. Refer to the documentation here.

Conclusion

Running open-source models locally with Ollama gives you full control over privacy, latency, and customization. With Clarifai Local Runners, you can expose these models via a public API without relying on centralized infrastructure. This setup makes it easy to plug local models into larger workflows or agentic systems, while keeping compute and data fully in your control. If you want to scale beyond your machine, check out Compute Orchestration to deploy models on dedicated GPU nodes.

MAICON brings together top visionaries and experts in the field of AI during a three-day conference packed with actionable sessions and networking events—all to position you as the change agent your organization (and career) needs. In this ongoing speaker series, we’re featuring these extraordinary leaders, with forward-looking predictions, actionable tips you can use today, and a preview of their MAICON 2025 sessions. Continue reading “How to Spark AI Adoption in Your Organization with Janette Roush [MAICON 2025 Speaker Series]”