← The Artificial Intelligence Show
The Artificial Intelligence Show
#179: OpenAI Government “Backstop” Controversy, Microsoft Humanist Superintelligence, Google’s Future of Learning, AI Driving Layoffs & Coca-Cola AI Ad Backlash
01:18:53
Loading summary
Transcript79 lines
- [00:00]
A
We have these two opposing sides. So there's the human should always remain in control Mustafa Salomon approach, and then there's the we won't have control and we should just accept that all like the all knowing AI is going to control us approach. So this is what's leading to increased chatter in political circles, religious circles and societal revolt is too strong of a word at the moment. Uneasiness, I'll say. Welcome to the Artificial Intelligence show, the podcast that helps your business grow smarter by making AI approachable and actionable. My name is Paul Raetzer. I'm the founder and CEO of SmartRx 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 Caput 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. Join us as we accelerate AI literacy for all. Welcome to episode 179 of the Artificial Intelligence Show. I'm your host Paul Raetzer along with my co host Mike Kaput. We are recording November 10, 11am Eastern Time. Could be an interesting week. I don't know, I mean I running out of weeks to launch all these models, Mike. There's all these rumors of models coming out and I think we still may see a few of them before the end of the year. And gosh, it's crazy to think we only have about what, seven weeks left in the year.
- [01:32]
B
Yeah, seven weeks. So you know, like two years in AI.
- [01:35]
A
Seriously? All right, well we, we have some really interesting big picture topics this week. I actually had, I don't know if fun is the right word, but certainly like intriguing process going through getting ready for this week's episode. Over the weekend there's some topics I'm just excited to talk about, I think that are going to open people's minds and sort of lay the groundwork for some of the bigger conversations I think we're going to start having about AI in society and education and business, things like that. So plenty to get into. This episode is brought to us by Mekon 2025 on demand. If you listen to the podcast regularly, you have heard us talk about this. So Macon was almost a month ago, right? Mike? Was it four weeks ago now? Kind of crazy to think about, but you can get on demand access immediately to 20 of the top breakout sessions and keynotes from that event. They're an incredible value for what you get. So it starts with my keynote, but you can also actually Watch my keynote on YouTube now. It's available for free on YouTube on our SmartRx YouTube channel. But we have becoming AI driven leader with Jeff woods, which is amazing. Mike's 38 tools shaping the future of marketing Better than Prompts with Andy Crestadena. Michelle Gansley, the former Chief officer for McDonald's on empowering teams in the age of AI. Human side of AI, which is incredible with, with leaders from the AI labs themselves talking about what's going on within the labs. Just an amazing collection of sessions. So check that out. You can go to Macon AI, that's M A I C O N AI and you can click on the On Demand bundle. We'll also put the link in the show notes and you can use AI Show 50 for $50 off of that. All right, so last week Mike, we introduced AI Pulse, which was our weekly now I guess we could call it a weekly series where we're going to do short surve surveys and try and kind of gauge what's going on with our audience, get feedback on key topics. So that's the premise is each week we're just going to kind of lead off with a couple of questions and then you'll have the week before the next podcast to go in and give us your thoughts. And they're quick hitting. These are not, this is not a demand gen thing for us. We're not even collecting email addresses as part of this survey. So it's literally just if you see your email address, it's only because it's tied to the Google form, which is we're using Google forms but we aren't, you know, connect Collecting emails and marketing to you as a result of this. It's literally just research for us. So we asked last week about following statements best describe your current personal feeling about AI's impact on job security. We had just over 100 responses. So this is a small sample size. This is not like projectable data. But it's more just like to give you a sense of kind of how our audience is thinking about things. So again, which do the following statements best describes your current personal feeling about AI's impact on job security? The number one answer, Mike, was it's a near term threat. I'm not worried about today, but I'm concerned about the next one to two years followed that was at 36.8%, followed closely at 28.3% with it's an immediate threat. I believe it's already causing significant job displacement. And then 27.4% said it's a long term rebalancing, it's going to cause some change, but you know, over time it's going to be okay. 7.5% said it's an opportunity, it'll create more jobs and value than it disrupts. And then no responses. It looks like Mike said it's overhyped. So no one that listens to our show feels it's overhyped. The second question we asked was which statement best describes your personal day to day use of AI tools in your professional work? By far the top answer was it's a habit, it's fully integrated. I use it daily as a core part of my workflow. That is 58.5% of respondents, 34% said they use it consistently multiple times a week for specific important tasks, 7.5% occasional, mostly for experimentation and non critical tasks. And then no one said rare to none, which doesn't surprise me. Right with our audience. And then as we said last week, we'll ask some questions just to get a sense of the audience a little bit more. So we'll ask things like job title, size of company, and again these are more just benchmarking to try and get a sense of like who the people are that are responding without collecting that data at a personal level and making it identifiable. So nice mix of audiences in terms of like industries, the biggest was 17.9% is professional services marketing. It's a lot of probably agencies, consultants I would guess, and then professional services. Other that was at 17.9, 14.2% was other, 13% in education. Like we knew we had a big education audience but it's nice to see that software was right there at 12%, manufacturing at 10%. So again, just to give you a sense and then the titles were all over the place. Yes, a lot of leadership titles like CEOs, founders, chief technology officers, chief operating officers, heavy dose of marketing and sales, directors of marketing, chief marketing officers, tech and data, software architect, ML engineer, data analyst and then product and strategy. Looks like it had a decent amount. And then consulting and education. So again we kind of sensed our audience was a pretty diverse mix of backgrounds of industries and titles and things. And this certainly seems to prove that out. So that was last week. Again, you can take part in this week's and we actually have all of these on the site. You'll be able to go back and look at this each week. So this week's we're asking, do you believe the concentration of power in a few major AI labs is a significant problem? And number two, is what should be the primary focus for advanced AI development right now. So both of these questions are going to make a lot more sense as we go through today's content, Mike. But again, you can go to what is it? Podcast SmartRx, AI forward slash, AI pulse.
- [07:21]
B
And we'll include a link to all of this and to this survey right in the show notes too.
- [07:26]
A
So, yeah, we'd love to have you participate. It's great to see these answers. And we'll keep this going again. That was the first one, so we had no idea what to expect if people are going to take the survey or not. So it's pretty cool. And we'll just kind of keep doing it this way. We'll share the insights from the week prior and then we'll put it in. And then we also will include the summaries of this in the weekly newsletter.
- [07:46]
B
And as an AI shout out here. Google Forms. Kudos, because Google has layered Gemini over Google Forms now. So all of these results were instantly summarized and they created the charts for me. And it was the work of just a moment to at a glance, see everything we needed to ask about when we were trying to figure out, okay, what were the responses? What's worth paying attention to? My gosh, like, I feel like we should use it for every survey moving forward.
- [08:11]
A
Yeah, that's what we did for the AI Council going into make every week when we had the council together. And I think I shared that on the podcast is it's one of those like, oh my gosh. Like life before having Gemini baked into forms and life after. It's incredible. And I've started to feel the same way with integration into Google Sheets. Like I use Gemini in Google Sheets all the time and it just seems like it's always been there. Like it's almost. I don't want to remember what working in sheets was like before I had it. All right. Well, we had, like I said, some big picture topics, one that was just kind of crazy to follow. Last week, Mike was starting off with this OpenAI and the blowback they got from what seems like a bit of a misspoken way of explaining what they were hoping to get from the government. And it led to some confusion and some retractions and. Yeah, and. But it actually is like a important topic on a bigger picture here about are we or are we not in this AI bubble? And so lead us off, Mike, with this OpenAI issue, I guess from last week.
- [09:09]
B
Yeah, they are facing a little bit of controversy right now because OpenAI CEO Sam Altman is basically pushing back on talk that the company company wants government help to fund its massive data center build out. Now this talk started because at a recent Wall street journal event, OpenAI CFO Sarah Fryer said all sorts of stuff. This was not the topic of the event, but in the course of some remarks during the event hinted that the government might, quote, backstop the guarantee that allows the financing, meaning the financing for data centers to happen. This set off some controversy because some industry watchers believed it meant that OpenAI wanted the government to essentially de risk or guarantee its trillion plus dollar bet on building the data centers and computing infrastructure needed to build advanced AI systems. This worried a fair amount of people, including the White House themselves. White House AI czar David Sacks posted on X there will be no federal bailout for AI. The US has at least five major frontier model companies. If one fails, others will take its place. A few hours later, Sam Altman put out a post on X addressing the controversy, saying unequivocally, OpenAI neither has nor wants guarantees for its data centers from the government. He emphasized that taxpayers should not be in a position where they might have to bail out companies that make bad business decisions. So Paul, you alluded to this. The reason this is happening and touching a nerve is because there's this deepening anxiety about the AI infrastructure race at large. Analysts have started to warn that this surge in spending from major players like OpenAI, Nvidia, Microsoft is inflating valuations and concentrating risk. And investors additionally are questioning how is OpenAI going to actually finance all these this $1.4 trillion in long term commitments to build data centers and computing infrastructure that is all needed to power the next wave of AI, given that it only projects 13 billion in revenue this year. So Paul, I guess reading this and kind of learning about the controversy, it came to mind like was this just a misunderstanding or a misinterpretation or did they accidentally say the quiet part out loud?
- [11:32]
A
Yeah, it's hard to tell. Sam's sort of retraction was a little bit wonky. It's like, hey, we don't want this but what we do want. And you could actually see by the but we do want how it could be misinterpreted of what they're actually looking for here. So I do think there's a lot more to the story of specifically what OpenAI is trying to do here. But I think if we zoom out, like why is this the lead topic today over potentially just like a misspoken thing from the cfo because I do think it gets at the uneasiness that investors are starting to feel, that economists are starting to feel that business leaders may be feeling around how impactful and important these like four to five companies are becoming. And so the key for me is that the US economy is actually becoming increasingly reliant on AI and the companies that are building and empowering it. So we're starting to see obviously the impact on jobs. We seem to talk about that every week. We'll talk about it again this week. And then GDP seems to be being affected by this build out of the data centers. So the spending on cap for energy and data centers by these big labs, if you think About Microsoft, Google, OpenAI, Meta XAI, we could throw Amazon in there. They're going to spend probably close to a half a trillion in 2026 and, and certainly trillions thereafter. OpenAI on their own plans to spend well over a trillion just in the next like six to seven years. So we're looking at trillions of dollars. Now. Why are they doing this? Well, because as we've talked about, the market opportunity is trillions of dollars. So the companies that build out these data centers, that control the cloud infrastructure, that serve up the intelligence that everyone is going to be demanding in every piece of software they use and every hardware device they use what I've called the age of omni intelligence, where literally the AI is just omnipresent in everything we use. And they're omni models, meaning they don't just do text, which doesn't require a massive amount of computer, they do agentic, they do reasoning, they do image and video generation. Eventually they will generate world models and video games. And so the compute demand is going to become so massive. The other thing tied to the economy is a lot of what we're seeing is these AI labs talking about the fact that the build out of energy and data centers is what's going to create jobs. So while jobs may be flat in other areas, they're saying, hey, we're going to hire 30,000 people over the next five years to build out all these data centers. And so truly the economy starts to become dependent upon this being true. So if what OpenAI is presenting as this future, and again not just OpenAI, these other major labs, they are all in on this build out and they need the energy, they need the data centers because they expect the demand for intelligence to keep skyrocketing. From a government perspective, they are very much on the record as saying they plan on, quote, unquote, winning this race against China at all costs. So the government needs these private companies to have these bold visions to take on enormous risks in order to get to superintelligence first, or AGI or whatever they want to call it. So the danger is that we become too reliant on these companies and they become kind of that quote, unquote, too big to fail. So the thing Sachs was referring to. So for context, what is too big to fail? Well, it was a book by Andrew Ross Sorkin in 2010 about the 2008 banking crisis. And then it became an HBO movie as well. So now what happened in the banking crisis? Many of our listeners are probably old enough to remember what was going on back in 2008. But in essence, the housing bubble and risky lending. So banks and mortgage lenders issued enormous numbers of subprime loans, which are mortgages to borrowers with poor credit, often with little verification of income or assets. These loans were then bundled into mortgage backed securities and collateralized debt obligations or CDOs that were then sold worldwide. The assumption of that economy was housing prices are just always going to rise. So this would never, we'd never have to like kind of pay the piper on this. So there was excessive leverage and complex financial products were created that like people didn't really understand. They started getting passed around and all of a sudden these banks run up in September 2008 of like, oh my gosh, they might fail. And what is going to happen if these big banks fail and they're not able to be supported privately anymore? Well, the government has to step in and intervene. So Treasury Secretary Henry Paulson, Fed Chair Ben Bernanke and New York Fed President Ted Geithner led emergency efforts orchestrating bailouts, forced mergers and eventually the 700 billion trouble asset relief program. So the government had to step in and fix the problem. And so that's where people are worried, like, oh my gosh, what if this happens again? What if we don't all understand the complexity of this infrastructure and all of a sudden OpenAI shows up and wants these unique financing packages put together. So there's an article the New York Times will link to that talks about this debt is under the AI boom. And so that article says to fund heavy spending on infrastructure for artificial intelligence, companies have leveraged a growing list of complex debt financing options. According to McKinsey, 7 trillion in data center investment will be required by 2030 to keep up with projected demand. Google, Meta, Microsoft and Amazon. So again, they're not even including OpenAI and XAI, which aren't publicly traded have spent 112 billion on capital expenditures in the past three months. To obtain the capital they need, hyperscalers have leveraged a growing list of complex debt financing options, including corporate debt, securitization markets, private financing and off balance sheet vehicles. That shift is fueling speculation that AI investments are turning into a game of musical chairs whose financial instruments are reminiscent of the 2008 financial crisis that we just talked about. Big tech companies are looking for new sources of funding. While Meta, Microsoft, Amazon and Google previously relied on their own cash flow to invest in data centers, more recently they've turned to loans. To diversify their debt, they're repackaging much of it as asset backed securities. About 13.3 billion in asset backed securities backed by data centers have been issued across 27 transactions this year, a 55% increase. So basically using the data centers as collateral to borrow money. Well, if the data centers demand for the data centers collapses at some point, all of a sudden the collateralization of the loans falls apart and somebody owes somebody else hundreds of billions of dollars. So you can see how this like, starts to almost feel like this shell game and it's like, oh boy, we better not be wrong. And so that's where the comments from the CFO come in, is they know the government needs them to do this build out. They're going to take that risk on, but if it doesn't go as planned, like, they don't want to be left high and dry. So the comments from Sarah Fryer, the CFO of OpenAI, she suggested the market is overly focused and anxious about this possible bubble and there isn't enough exuberance going on. But then she talked about we're just building out full infrastructure today that allows more compute to come into the world. I don't view it as circular at all when she's talking about, like how this is all being done and these deals between. But then she said in addition to OpenAI's deals with chip makers, the ChatGPT maker is also eyeing a broad mix of financing vehicles to fund its infrastructure. Fryer said OpenAI is, quote, unquote, looking for an ecosystem of banks and private equity to support its ambitious plans. And then this is where all the problem came in. She also hinted at a role for the U.S. government to, quote, backstop the guarantee that allows the financing to happen. So that's where like all this blew up is the word backstop. So she then actually had to retract this and post it on LinkedIn. She said, I want to clarify my comments earlier today, OpenAI is not seeking a government backstop for our infrastructure commitments. I used the word backstop and it muddled or muddied the point, as the full clip of my answer shows. I was making the point that American strength in technology will come from building real industrial capacity, which requires the private sector and government playing their part. As I said, the US Government has been incredibly forward leaning and has really understood that AI is a national strategic asset. So long story short, AI is becoming increasingly political and its impact on the economy is growing way more than most investors or business leaders realize or understand. And so no matter how they try and clarify this, the reality is you have private companies taking on enormous risks that the government is encouraging them to do and needs them to do. And of course they're going to try and find ways to say, okay, we're going to do this, but we, we need your help. If, if at some point like this isn't going as planned. And David Sachs, again, who's the, you know, the AI guru for the administration, speaks on behalf of the administration, is like, yeah, tough luck, like you go under, we got four other labs we're depending on, so you're on your own. But then everyone hedges. So even Sachs is like the tough guy. Like, oh, and then he comes back and what, his, his last thing. I think I had it in here. Yeah, he said finally to give benefit of the doubt, I don't think anyone was actually asking for a bailout. That would be ridiculous. But company executives can clarify their own comments. So you know, again, everyone's going to start talking about this, but if there's a, this much conversation going, going on, people may use the wrong words here and there, but they absolutely are going to want the support of the government to take on the risk they're taking on, however you want to phrase that or whatever that looks like.
- [21:19]
B
And so just to be sure I have this clear, the real risk here, the real danger is there are all these pie in the sky projections about the demand for intelligence moving forward. If there is not, if that demand does not materialize, given the speed and scale of these data center buildouts and investments, these companies can be left high and dry being unable to cover their loan obligations because not enough people are using their products.
- [21:45]
A
Correct. The economy's screwed.
- [21:46]
B
Right?
- [21:46]
A
Yeah. So like who's the, who's the dude, the big short? Who is.
- [21:50]
B
Yeah, yeah, Michael, Michael Burry.
- [21:53]
A
Who is the.
- [21:54]
B
Michael Lewis?
- [21:54]
A
Read the book.
- [21:55]
B
Burry was the guy who made the big short on the company.
- [21:59]
A
I Believe I just saw last week, he took a billion dollar position against this build out. So the guy who bet against those mortgage backed loans is now betting against the build out of AI. But yeah, Mike, you're 100% right. The assumption in all of this is that scaling laws continue. The models keep getting smarter. They increasingly can do human labor. So we talk about this $11 trillion annual wages market within the US so the assumption is they will continue to be able to do more and more of that work and that we humans will continue to demand an insatiable demand for intelligence in every product we use and every piece of software we use. If that holds true, then all these data centers, all this energy we're trying to create will be used and we will probably still be at a deficit for it. If at some point supply and demand gets out of whack, we're screwed. It's basically where this is all going.
- [22:59]
B
All right, that is a good clear breakdown there. I love that. So our second big topic today concerns Microsoft. So Microsoft is actually forming a new super intelligence team. And it's got a bit of a twist. And the twist is they promise to keep humans firmly in control. So the head of AI at Microsoft, Mustafa Suleiman, who we've talked about quite often, published an announcement called Towards Humanist Superintelligence, where he said this new team will pursue what they call humanist superintelligence, or what they would call powerful AI systems that are explicitly designed to serve, not surpass humanity. Suleiman argues that the world's current race towards AGI misses a deeper question, which is what kind of AI should humanity actually want? And their answer is instead of unbounded autonomous systems that outthink humans, Microsoft is proposing a model that is built more for containment, alignment and purpose in serving humanity. So Humanist Superintelligence, or hsi, envisions essentially domain specific systems that solve concrete global challenges, things like medical diagnoses or clean energy, without drifting toward uncontrollable behavior. So he describes it as super intelligence with limits, which is designed to keep humanity in the driver's seat while amplifying progress. So this approach kind of rejects both the doomer fears and the accelerationists, positioning Microsoft's new team as a long term steward of safe and practical AI. And Suleman even wrote in here, humans matter more than AI. So Paul, I'm curious, why does this manifesto or this kind of positioning really matter here?
- [24:44]
A
So I don't think the timing is a coincidence with what's going on with these super intelligence labs at meta and now OpenAI very openly talking about superintelligence. But the timing was particularly keen based on what happened with Elon Musk last week. So in essence what's going on is we have these two opposing sides. So there's the human should always remain in control Mustafa Solomon approach. And then there's the we won't have control and we should just accept that like the all knowing AI is going to control us approach. Believe it or not, like if, if that sounds weird to you, there is an absolutely techno optimist cam who assume it is inevitable that AI takes control and that that's okay, that we as a species, like eventually other species come along that are just more powerful and that's who ends up leading society. So this is what's leading to increased chatter in political circles, religious circles and societal revolt is too strong of a word at the moment. Uneasiness. I'll say. So if you follow Twitter closely, there is. You can't go a day without leading politicians talking about AI now. And as a matter of fact, the Pope is like a regular tweeter now about AI. So we are going to see this increasing, I don't know, like division within society of where does this really all go. So to the we should cede control to the all knowing AI. We'll get into Elon Musk in last week's Tesla shareholder meeting. So if you didn't follow this story, Elon sort of got a major pay package worth billions of dollars taken from him because it was determined to be unlawful was how the package works. So basically Elon threatened last year to leave Tesla in not so many words, if they didn't approve a new pay package for him. That was not so much about the money. He's already the richest person in the world. It was about control and control when humanoid robots become available at scale. So in essence, what was happening is Elon sees the future of Tesla as a humanoid robot company. The cars are going to be very secondary to what he thinks the overall market opportunity for Tesla is. And so he's envisioning the AI he's building at Xai his lab being embodied within Tesla robots, which he claims will be the biggest product in human history because everyone will want to have probably multiple humanoid robots. So billions of humanoid robots will exist is what he thinks is going to happen. So he wanted a $1 trillion pay package approved. 11 trillion for an individual person. But the more important thing was that 1 trillion represented control of Tesla and control of where these robots go. So again, this is A bit of a side story, but it's interesting. So CNBC has a breakdown of how this works, this trillion dollar pay package, which sounds absurd, but we'll run with it for a minute. So the pay package for Musk, already the world's richest person, consists of 12 tranches of shares to be granted if Tesla hits certain milestones over the next decade. It would also give Musk increasing voting power over the company, acceding to demands that he's made publicly since early 2024. His ownership would increase from about 13% to 25%, adding more than 423 million shares. So again, this is not about money to him. This is about control. The first tranche of stock gets paid out if Tesla hits a market cap of 2 trillion, which they will probably hit by spring of next year. Tesla's current market cap is 1.54 trillion. Awards tied to market cap gains are paired with operational achievements. The next nine tranches would be awarded as the value of Tesla's market cap increases every 500 billion. So every half a trillion, he gets another tranche of this, the, these shares. So up to 6.5 trillion, he would earn the last two tranches. If the market cap rises by increments of 1 trillion, meaning when they hit 8.5 trillion, so they're 1.5 today. When they get to 8.5, then he would get the full package of a trillion dollars. So that seems crazy, but that is the backdrop to last week's shareholder mealing where this pay package was approved. So this is now moving forward. It's the comments he made, though, during the shareholder meeting that leads to why we're connecting this to the Mustafa thing. So here's what Elon said. People often talk about eliminating poverty, giving everyone amazing medical care. There's actually only one way to do that, and that's the Optimus robot. Musk later doubled down. Optimus will actually eliminate poverty. The robots would increase the global economy by a factor of 10, Musk said, or possibly even 100. Optimus Robot will have five times the productivity of a human per year, he predicted, because it would be able to operate 24. 7. And this is a direct quote. I came to the conclusion that the only way, the only way to get us out of the debt crisis and to prevent America from going bankrupt is AI and robotics. He then said AI and robots will replace all jobs. Working will be optional, like growing your own vegetables instead of buying them from the store. But the one that directly leads into the Mustafa conversation, he gets asked a question by A guy in the audience, you can watch this video for yourself. He said, long term, the AI is going to be in charge, to be totally frank, not humans. If AI vastly exceeds the sum of human intelligence, it is difficult to imagine that any humans will actually be in charge. So we just need to make sure that the AI is friendly. So that was Elon. Last week, Mustafa's tweet tied to this article he wrote, said it shouldn't be controversial to say AI should always remain in human control, that we humans should remain at the top of the food chain. That means we need to start getting serious about guardrails now, before superintelligence is too advanced for us to impose them. So that's what I think. Again, they've been working on all this, but it's not like Elon's perspective is new. It's just point blank saying it now. But this has been talked about for a while. And so it's interesting because now you have Microsoft, which I don't know. I'm trying to think about this for a second. I'm trying to, like, this is probably the first major ad company to come out and just directly say this. Like, Anthropic has sort of said some similar things, but I also feel like Dario is just sort of also under the assumption, like, yeah, the AI is just probably going to be in control at some point. So this is a very different position than the other labs have taken, I would say. I'm trying to think of. Google has directly addressed this.
- [31:41]
B
Well, to that point, too. Have you heard of anyone? I think he called out in here. I'm trying to find the exact language he called out. Basically, like, we're okay with fewer capabilities if it remains in human control. Like, have you heard of anyone say, like, well, we're just going to not make it as smart as it can be so that we can remain in control? That seems to me.
- [32:00]
A
Yeah, that's the thing is, like, it's one of the parts that keeps being brought up as, hey, at some point we may have to, like, come to an agreement as labs to not push forward. So it was talked about, you know, with OpenAI and Sam even said this recently. I've certainly heard Shane Legg and Demis talk about this at Google, DeepMind, that, like, there might come a point where we all have to work together. Like, if someone proves that these things truly are becoming dangerous and they're past our ability to align them, that there may be this reckoning where we all have to get in a room and say, you Know what? We do need to pause this. Like, we, we are now afraid of what we've created. So they have talked about that. To have someone like Mustafa who leads AI at Microsoft basically say that they're willing to put the brakes on when everybody else keeps accelerating. That might be the first time I've seen that where like a leading a leader at a lab is saying, hey, listen, we're kind of getting to that point and we need to define this as human humanist superintelligence. We're going to give it a name. We are going to adjust our product roadmap and our own company strategy. We're going to keep spending these tens of billions, hundreds of billions. We're going to start building our own models. We're going to do also. But like we're going to do it specifically to solve problems and make them domain specific. They're not going to be unbounded. They're not going to just be able to do this runaway intelligence thing. Containment is necessary. Like all this language. It's the first time I'm seeing that from like a lab that's saying we're going to be willing to do this differently. Now my whole takeaway, because again, he kind of wraps it with their big thing is like AI companion for everyone. Which is weird wording, but like that everyone should have a perfect cheap AI companion that helps you learn, act and be productive and feel supported. Medical super intelligence and then plentiful clean energy. That was kind of a weird three things. Honestly. I was like, that's. I don't know. But that was how they kind of ended it. Here's where I keep coming back to though. Mike is like, so Mustafa is a Google DeepMind co founder, spent time at Google DeepMind went and did inflection with Reid Hoffman, who is not in the techno optimist, like race ahead and build this stuff. So he's more similarly aligned. They did. What was the book called? Super Intelligence wasn't. Or Super Agency.
- [34:17]
B
What was the Super Agency? That sounds. I think that's Hoffman's most recent.
- [34:21]
A
All right, you can look that up real quick while we're doing this. But. So he came up with a book last year where they basically wrote about everything he's saying here. The thing I can't like, I keep trying to make this add up. I just don't know that Mustafa realizes this vision at Microsoft. At some point, Microsoft has to compete with all these other companies. Like Microsoft's the second most valuable company in the world, give or take. I think it bounces between second and third, if all the other labs keep racing forward and pushing superintelligence and they're willing to commercialize it in ways that Microsoft isn't, at some point there's a fiduciary responsibility to the shareholders to maybe not always follow through on this humanist superintelligence position. And so I can't help but feel like eventually this clashes with what Microsoft has to do to justify their investment in AI.
- [35:20]
B
Right.
- [35:20]
A
I may be totally wrong on that. And maybe Satya and everybody else at Microsoft is like 100% lockstep with Mustafa on this, but when I read stuff like this, I generally feel like it's Mustafa's vision and belief and not always 100% aligned with, like, what Microsoft is actually going to do, I want to believe it like, this is actually the most aligned position I personally would have. Not that my personal, like, alignment matters that much here, but this is like, I feel like there just has to be this middle ground. I've always felt that, but I feel that in everything. Like, I feel that in politics, I feel it like AI. Like, I always just feel like, why can't we just rationally listen to all these sides and like, let's arrive at what a reasonable middle ground is. It doesn't have to be. AI is going to take over and let's just give in. It doesn't have to be. We have to stop everything to figure this out. It's like, no, like, what's the center part? Like, can we just all get in a room and actually find a reasonable way to do this? So I don't know. That's the one thing I keep taking away from this. But it's a good read. Like, if people should go read it, I just don't know how much it's going to actually come to life.
- [36:25]
B
Well, that is why all these numbers we often talk about matter so much. It's like we talked last week about how Microsoft had to take billions of dollars of write down on its investment in OpenAI because they're not making any money yet. It's like money and everything, but that's what they're incentivized to optimize for. So.
- [36:43]
A
Correct.
- [36:43]
B
If they see a future where it's like, ah, we can 10x this investment or like make good on it, but it requires Sam Altman going full bore on AI or super intelligence, then you get into conflict.
- [36:55]
A
Right? Yeah. And I. So even now I'm thinking out loud again. Microsoft acquiring Ilya Sutskova and safe superintelligence becomes Much more interesting thing I hadn't really thought about before.
- [37:09]
B
Yeah.
- [37:09]
A
So Ilya obviously isn't a Microsoft guy. You know, came from the Google tree as well and then into OpenAI. But like safe superintelligence certainly aligns with this humanist superintelligence. So if you start looking at who's aligned, it's anthropic. Ish. So if I put by the bucket these, like, Microsoft is now kind of like in that certainly safe superintelligence, a little bit anthropic mentality. Google DeepMind I still believe is truly about like the, the good of humanity and the scientific discovery. But like, I'm not sure yet how much they'll push super intelligence before they would like arrive at a point like this where they would write this kind of essay. But they, they own the, you know, the biggest share of anthropic, I think at like 14% or more. I don't know. Like, I, I still think there's consolidation of labs here at some point. And it's. So it's just interesting to start to think about how the different labs look at this big idea of like, what do we do when super intelligence starts becoming real or within reach?
- [38:07]
B
Well, now that you mention it, I could be convinced of an argument that this is just fishing for Ilia or the next Ilia. Right. Like, this is how you get talent. Not only the billions of dollars people are spending on talent, but the perspectives on the technology do really matter to these researchers.
- [38:24]
A
Like, we're not going to pay you what Zuck's going to pay you at Meta, but.
- [38:28]
B
Right. Come build the actually good super intelligence instead of whatever they're doing.
- [38:34]
A
Yeah, it's the human centered thing. Yeah.
- [38:37]
B
All right. Our big third topic this week is that Google has published a paper that details more about how it's planning to integrate AI into global education. Now, this paper is titled AI and the Future of Learning. And in it, Google leaders outline how AI could help address a worldwide teacher shortage and declining academic outcomes while tailoring learning to individual needs. And they argue AI's role in classrooms should be to amplify human teaching, not automate it. So they outline all sorts of ways that AI is going to have an impact in their, from their perspective on the future of education. So a couple things that jump out in this paper. First, they outlined how their Gemini AI models are now grounded in what they call core learning science through their system called Learn lm. And this basically embeds proven pedagogical principles into the models. And in their words, this makes Gemini the world's leading model for learning. So Gemini can now act as a tutor, lab, partner, study coach across different tools like search, YouTube and classroom. According to Google data, Gemini outperforms other models on every learning, science, metric. And second, they kind of prescribe that AI driven personalization at scale in education needs to be the norm. We need tools that adapt lessons, feedback and pacing to each student, while freeing teachers to focus on creativity and mentorship. They also call for new assessment models that reward reasoning, collaboration and originality. Google actually argues that schools should maybe move towards oral exams, project portfolios and debates, which are areas AI can't easily simulate. And they also prescribe some systemic safeguards, so things like adding in safety filters, privacy protections for under 18 users, and red teaming specifically for educational context. So, Paul, the impression I got here was that Google's really giving some serious thought to how AI is going to reshape education as we know it. I mean, it's certainly just the initial flag planted in the ground here, but it seemed, I'd be curious your perspective, especially building an education platform, how you're looking at what they're proposing and where this is going.
- [40:52]
A
Yeah, so the reason I wanted to do this one as a main topic was a few things. The first one, I read it over the weekend, I'd flagged it last week and then I read over the weekend and it's nothing like newsworthy in terms of some major announcements they're making or some major shift. But I did read it from the perspective. Okay, we're trying to build an elearning platform and how is this going to affect professional education? How would that trickle into what doing with AI Academy by SmartRx. And so there was that, but then there's, you know, I'm a parent of seventh and eighth graders and so I'm constantly battling with this integration of AI into education. I'm looking at what the schools are doing where, you know, my daughter will go into high school next year. I'm assessing the high schools based on what their AI roadmap is and what their policies are. I talk to family and friends all the time who have kids at different levels, including higher education, where they're struggling with like it's not even integrated, my kid's not learning anything, like what, what are they going to do for a job in a year when they come out? And then as we highlighted in the AI Pulse up front, we have a lot of educators in our community, a lot of educators in our audience who are trying to drive innovation and AI Adoption. So I think it's interesting to, from all those perspectives to look at what is Google saying and what are they talking about. And so, you know, I think when I was going through this, I'm trying to just get a hint of where are they going to go with this? Like, where's the product going to go? How is it going to evolve? Like my kids have Google classroom, that's how all their assignments are done, that's how they submit everything. They use Google Docs, they use sheets. And I was actually working with my son yesterday, on Sunday with him and a friend on a pitch competition they have to do. And so we were in Google Slides and then I was actually, they'd done a survey and they went over to like, you know, exported into sheets and there's no Gemini in there. There's no like guided learning capability in the, in the classroom feature of Google. And I was like, ah, I wonder when that's coming. Because I was actually hoping it was there because I was going to show them how to talk to Gemini about the survey data, but it couldn't do it because it wasn't embedded. So I think it's just really interesting and so some of the things you highlighted, Mike, but I'll just call out a couple of the things I had bold faced as I was going through this. So they talk about AI unlocking human potential all around the world and tying to their overall mission of making the information universally accessible and useful. AI does present urgent challenges and unknowns that society must collectively reckon with. Their goal is to help improve learning outcomes by developing AI products that are grounded in that core learning science you mentioned. And in close partnership with education community, they talked about learning being the bedrock of human potential and societal progress, equipping people with new skills, sparking curiosity, exchanging fresh ideas. They did get into today's jobs demand not just these foundational skills, but advanced problem solving, collaboration, the ability to learn throughout life. And like that. That's interesting. He's like, well, how do we make sure they keep doing that, that students and even young professionals continue to have that curiosity. I was, oh yeah, okay. I don't, I don't want to get into like naming names here. I was having a conversation recently where someone said something about like, well, how are you doing school? And they're like, oh, I literally just use chat GPT for every assignment. Like, and this is like at a high school level. And, and so the, the friend was like, well you can't do that. Like you're not learning anything. And they're like, yeah, whatever. My assumption is there's probably a lot of kids who are like that. Like, as long as they can get away with it, they're just going to keep doing it. So, yeah, I don't know. I thought there was just a lot of interesting perspectives. So for educators, I would definitely read this. For parents, I think it's really good to be aware of how the tech companies are thinking about this and what they find to be important and the different challenges that they're currently solving for. They did end with some interesting questions, like things that they're thinking about how AI will change what we need to learn, or even what it means to learn. How might historical forms of evaluations and assessments change due to AI? How will the nature of teaching evolve? And how can AI facilitate new types of learning previously not possible? So I liked all of those ideas and then one related note. Mike I was listening to Dwarkesh podcast with Andrej Karpathy recently. I don't think I mentioned this on the podcast yet. I mean, Andre's is always amazing to listen to, but he's working on like the future of education himself. And so there was this part that I remember, like I stopped when I was listening to it and like rewound it and like, listen to it again. And so they were talking about this assumption that, like, AGI does happen, that we get to the stage where the AI is at or above human level, at least average human level, at basically all cognitive tasks. And so he said to Andre, he's like, well, what happens then? Like, if we get to AGI, like, do we still need school? Do we still need education? Like, what does that look like? And he said, I often say that pre AGI education is useful. Post AGI education is fun. And in a similar way as people. For example, people go to the gym today, but we don't need physical strength to manipulate heavy objects because we have machines for that. They go to the gym. Like, why do they still go to the gym? It's because it's fun, it's healthy, and you look hot when you have a six pack. What I'm saying is it's attractive for people to do that in a certain very deep psychological, evolutionary sense for humanity. And so I kind of think that education will play out in the same way you'll go to school like you go to the gym. That's so interesting because I think there's people who are just naturally curious, lifelong learners who will always seek education, they will always take courses, they will always go back to school and get another degree or just take some interesting. Or watch a YouTube channel about something they want to learn because it makes them feel fulfilled and good. And that's kind of what Andres is saying. He's like, maybe that's what education becomes is like, it's just for the people who want to keep learning. And if you want to, you know, kind of peak as a human and let the AI do everything, like, you'll probably do that. Just like you probably won't go to the gym and eat healthy and stuff, like, so that. I don't know, I just thought it was like, a really interesting perspective, sort of a related topic, of the future of education.
- [46:46]
B
Well, you know, what's interesting is during that podcast as well, he mentioned something to the effect of, look like the times I have learned the best have been when I've had, like, a direct tutor helping me work through problems at the right pace for me. And they mentioned in the Google paper, they even say, as part of their aspirations, ideally, every student would spend ample time working their zone of proximal development, they call it, which is the sweet spot of just right learning. Challenges that lead to new skill growth. And, like, no knock on the education system. But that's, like, not very possible today. So if you're thinking of it that way, you're almost like, we've been teaching with two hands behind our backs for 200 years. Right. So. Or longer. So that's where the real excitement, I think, comes in. Can I make possible that personalized tutoring or learning, like, we might all be able to learn far better than we ever have.
- [47:40]
A
Yeah. And I've definitely spent time with a lot of educators who see this opportunity, who know education is nowhere near reaching its potential to impact people.
- [47:49]
B
Yeah.
- [47:49]
A
Because it can't be personalized and it can't be adapted based on the way people learn. That's why things like Khan Academy and Duolingo have been so popular, is it enables that kind of learning. And that's like, what we're trying to do with Academy is like, reimagine the way to do this stuff. So, yeah, it's just. I mean, education touches all of us. If it's not you personally, it's your family, it's your friends, it's, you know, your co workers, it's your employees, if you're a leader, like, and so thinking about this, how we do reskilling and upskilling in a professional environment, how we guide our kids at different levels of education, like, this is just a fundamental, critical topic as we move forward. So we're going to, you know, try to do our best to start making it just a more regular part of what we discuss.
- [48:29]
B
Yeah, for sure. And if you needed any reassurance why this is such an important topic, our first rapid fire is definitely tangentially related to reskilling and upskilling because unfortunately, the US companies announced more than 153,000 job cuts in October, and this was the highest for this month in over two decades. And according to some new data from Challenger, Gray and Christmas, they show that AI is a factor here. So they show that layoffs nearly tripled from the rate a year ago. And the firm's Chief Revenue officer, Andy Challenger said the cuts reflect AI adoption, softening consumer and corporate spending and rising costs, which is driving companies to tighten belts and freeze hiring. Now, we've talked about some of these layoffs. We have Amazon, Target, Paramount have announced sweeping reductions. Each of them have cited automation and management restructuring. We talked about ups eliminated 34,000 and operational roles. And year to date job cuts have topped 1 million, while hiring plans are apparently at their lowest since 2011. So some employers like JP Morgan say they'll redeploy workers affected by AI rather than reduce headcount. But it sounds like based on this data, for many, finding new roles is getting harder. So, you know, they call this out, Paul, explicitly in the data they call AI out. They say October's pace of job cutting was much higher than average for the month. Some industries are correcting after the hiring boom of the pandemic. But this comes as AI adoption softening consumer and corporate spending and rising costs drive belt tightening and hiring freezes. So seems like we've talked about this issue at length, but it does seem more and more like people are pointing to AI at least as one factor.
- [50:11]
A
Yeah, it is. It's coming to this recurring topic every week, unfortunately. And I don't know, like, I really want to be optimistic here, but I've been pretty consistent saying I think there's going to be some short term pain. I don't know how short term that is. Like, I don't know how long that short term pain lasts. But I mean, I had some conversations last week. I was at a couple of major events and there, there's more coming. Like there's, there's ways to know that cuts are going to be occurring that the general public and the economy isn't hearing about yet. And I will just say, like, there's quite a, there's some indicators that we're just Sort of at the leading edge of this. And it's, it's in a way imminent that the next, like three to six months, we're probably going to see some pretty significant cuts that may, people may be a little bit more transparent about their connection to AI, but I have a relatively high level of confidence that we need to be having these conversations and preparing more for continued cuts that are going to be affected in at least part by AI. So, yeah, I mean, like we always say on this podcast, we have to be realistic about this. Like, ignoring it's not going to do anything. And I think we're probably past the point where we can just deny that AI is going to have any impact on jobs in the economy. I mean, hopefully people have kind of accepted that by now. We just have to move as quickly as we can to, like, what do we do about it? Like, we have to just, just be more proactive about this. It's not going to stop anytime soon.
- [51:52]
B
Yeah. And in case people think we're harping on it, I mean, I don't think either of us get any enjoyment from having talked about this, but, like, look at that AI pulse survey. It's not just us. 65% of the audience either thinks today it's a near term existential problem or in the next one to two years, and they're not just taking that from us.
- [52:09]
A
So, yeah, I would happily lead every episode saying AI is creating jobs and here's where it is and here's what the data is telling. It's like, trust me, I am anxiously awaiting the day when we can switch gears and start talking about the growth and innovation and jobs that are being created. That is not currently in sight. Like, there's isolated, certainly it's creating new roles. Like, we talk about this all the time. Like there are new roles that are being created. They're just nowhere near at the level at which they're going to disappear in the coming months.
- [52:44]
B
All right, our next rapid fire topic. A year after its first AI Generated holiday commercial drew backlash, Coca Cola is trying again. And they say that this time they've gotten it right, or even more right from their perspective. So the company's new seasonal ad is once again AI Generated. It mimics the style and tone of past Coca Cola commercials. It's got some cartoony animals watching the brand's famous red trucks travel through the snow. They have a reveal of Santa Claus stepping out of one of the trucks. And this was produced by the LA based AI Studio Secret Level. And the new commercial was built Almost entirely with generative AI tools. Multiple models helped shape the concept, visuals and the animation style. Every frame was generated and refined through AI prompts guided by a small team of human creatives. And Coke's Global VP for Generative AI says that the craftsmanship on this ad is 10 times better than the previous ad. There are advances that now allow for more natural emotion and movement in the videos. And if you recall, last year's ad drew controversy for using AI instead of more people to generate the company's kind of iconic Christmas creative. And this time, no different. Viewers have criticized the visuals. They've questioned the trade off between speed, cost and artistic value. Critics have raised environmental and labor concerns, warning that AI driven campaigns could accelerate job displacement across the creative industry. So, Paul, it seems like it is becoming a holiday tradition for people to get pissed at Coca Cola about their AI ads. That doesn't seem to be changing. It is interesting. In a couple of the articles here, the Hollywood Reporter has some really good in depth dives with Coke and with the agency about how they made these super fascinating, but they're pretty unapologetic.
- [54:32]
A
I think they have to be like, I think you just have to own this. The way that I think this plays out is people who watch this ad and don't know they generated and don't like have strong feelings about AI one way or the other, probably think that's a really good ad. That was beautiful. It put me in the holiday spirit. And there's going to be people who watch it, who knows AI generated, know the backstory and they either like or don't like AI. They either are the, what was the training data? You know, reply boys. Or just like every single AI thing is like, what was the training day? What was the training day? Which I get like, I understand if that's the talking point. I, I just feel like Coke is gonna go out into that frontier, they're gonna piss off 40% of the population and, and then at some point I was going to just stop being pissed and they're just going to accept that this has evolved creativity.
- [55:23]
B
Yep.
- [55:24]
A
And somebody's got to do it. Like people have to go out into that frontier and be willing to be different and do what they consider to be an evolved form of creativity and they're gonna get crap for it. Like, I don't, I don't know that I saw a single positive post on X about this. Like, yeah, X is definitely a bubble of people who are, are challenging this evolution. And again, this isn't me taking a personal Stance on like saying, just get over it and let's move on with life. I totally get it. Again, I've said this 100 times. My wife is an artist, my daughter's an artist, I'm a writer by trade. Like, like, I get it that, that this is not a black and white, like zero. And one thing, it's not a binary decision. It is like there's a fuzzy middle ground here. There's things that are, are messy and uncomfortable. But brands are going to keep moving. I think at some point society just sort of like, it just becomes creativity. Like it's, it's different and it's going to take a little while. But I don't know, I mean, good for Coke, for standing on the brand like this is what we're doing and they're not backing down.
- [56:31]
B
So yeah, I think this is one of those areas too. I think you just have, and rightly so, so much emotion and personality and identity tied up in some of the skills here and the creativity of it and the humanness of it that this is where you start to see some of that backlash in this area especially.
- [56:48]
A
Yeah. And it's take. You can't. Like, that's the thing is like the creatives are using the tools available to them and they're doing incredible things with those tools. Is it their fault that they're trained on data that was stolen? Like, are they just not supposed to use the tools? Like this, Is that weird? There's no perfect answer. There's what people believe. Like there's these subjective opinions about was it right or was it wrong and things like that. Versus, you know, I think there's a lot of people, which I probably put myself in this bucket of like, it just is like, I can't change how they train the models. I can choose as a company, are we going to use the models or aren't we? And we choose to use the model. So I suppose, like, in a way I'm saying, well, if I don't believe they should be allowed to do this, then we probably shouldn't be using ChatGPT and Google Gemini right in our company. And that doesn't seem like a sustainable business decision. So it's hard. Like, I get why people would feel strongly on both sides of this.
- [57:46]
B
All right, next up, Amazon has filed a federal lawsuit against Perplexity, accusing the startup of illegally using its new AI agent to shop on behalf of users. So at the center of this dispute is Comet, Perplexity's AI browser, which can log into users Amazon account Search for products and complete purchases automatically. Amazon says this violates its terms of service. The company says Comet disguises itself as a normal Chrome user, which degrades the shopping experience and creates privacy risks. Perplexity CEO argues instead that agents should have the same rights and responsibilities as human users acting on their own behalf and called Amazon suit a bully tactic to suppress competition. So I'm curious about the nuances you see here, Paul, because it doesn't sound like Amazon, from what I was reading, is totally against agents. But their CEO Andy Jassy did say on an earnings call last week that the current customer experience for AI shopping agents was, quote, not good. He cited a lack of personalization and user specific shopping history and bungled delivery estimates and pricing. But he did seem to hint at he thinks there's going to be ways they find to partner with companies related to agents. So what do you think here?
- [59:00]
A
I don't think they have a choice. I mean, this is the future. Agents are going to be able to do shopping. So obviously Amazon has to have a play here. This is a tricky one for me. Like Aravind, the CEO has a very publicly available. You can go look at it yourself. Questionable legal and ethical choices. So I've referenced this before, like, brags openly on podcasts about the fact that he built a company that scraped LinkedIn data knowing it was against the terms of use, but everyone else was doing it, so he justifies it by everybody else doing it. So again, it's like I have a hard time feeling empathy for perplexity who knowingly abuses rules in terms of use. So anytime that they're talking about, like, oh, we're, you know, the victim here. It's like, come on, like, you guys have made an entire company worth billions of dollars on stealing from people. Like, this is what you do. So take that for what it's worth. Do I think Amazon is in the right? I don't know. Like, I'm not a lawyer. I didn't analyze like exactly what the lawsuit is here. I went through like the basics of it, but this bullying is not innovation thing I just kind of like laughed at. But regardless, I'll pull a couple excerpts. So this is the article from Perplexity that like when they got this lawsuit served to them the point of technology is to make life better for people. We call it innovation, but it's just the content constant process of asking how to make things better. Bullying, on the other hand, is when large corporations use legal threats and intimidation to block innovation and make life worse for People, AKA when we steal and we break rules, they, they try and stop us from doing it. And that's bullying is basically what this is. So like the rest of it is just like, okay, whatever, like, like then they go into how this is just the future and Amazon should just accept it and agents are going to do this stuff and which is probably right. Like they probably will. Like, this is where it's going to go. Doesn't mean that you just get to rewrite the rules. And so this is like this approach from, from technology overall. It's like, well, we're just going to steal the 7 million books because we think we should be allowed to. Even though it's against the law. It's like, okay, but that's not how the law works. That's not how terms of use work. Like there are standards in society that we're kind of supposed to abide by, but that it just isn't the techno optimist, like accelerated all cost position. So again, I'm not trying to make like, you know, who's, who's the villain, who's the hero here? I'm not, I'm not trying to play that role. But like, don't just look at this like bullying, stopping innovation, whatever, and like on the surface and think, oh my God, perplexity is right. Like, Amazon's the bully. It's like, no, there is way more to the story here. That being said, when I think to the future and we're probably going to do like a 2026, like AI Trends special episode. We'll talk a little bit about this, but one of the ones, Mike, that I had kind of jotted to myself is this agent to agent communications and commerce. It's like a couple months ago when I was thinking about this where I said businesses must solve for consumers, using agents to gather information, engage with brands and make purchases. This may alter how we design user experiences on the web and in apps and it could rapidly evolve marketing, sales and customer experience strategies. So this is like, like we're all going to face this. This is Amazon doing it now. But like, what's stopping an agent coming to your brand's website and interacting with your chatbot and like annoying you or going through processes where you're losing insights into like the analytics and you have no idea what users are. Like this is going to happen to all of us at kind of like a micro level. So it's, it's a, I guess a good topic from that perspective.
- [62:25]
B
Yeah, it seems like such a fundamental question.
- [62:28]
A
Right?
- [62:28]
B
Because I just can't help but see agent to agent, like, like rewriting every assumption of marketing and of E commerce because it's all based on the brand at some level, mediating the experience. And that goes away with agents.
- [62:43]
A
And I haven't really heard of a brand that's figured this out yet, in part because it, it requires a leap of assumption about the agents becoming more autonomous and reliable, which they aren't right now.
- [62:52]
B
Right.
- [62:52]
A
And so, but that doesn't mean they won't be by Christmas season, by like holiday season, 2026, like 12 months from now. This may be a very real thing affecting like every website of our listeners. So it's like we're almost sort of that I would say like trying to see around the corner, like three, six, 12 months. Yeah, this is one of those where like we're kind of looking around the corner at the moment and it's going to become very real to a lot of other people in the near future.
- [63:18]
B
Next up, we have some pretty juicy gossip, some inside baseball here in the world of AI. A newly released court deposition has shed light on how close OpenAI once came to Mergin with its top rival Anthropic, and how some of the internal power struggles played out when they tried to oust Sam as CEO at OpenAI. So this testimony actually comes from OpenAI co founder and former chief scientist Ilya Sutskeva, who spoke under oath in a case tied to Elon Musk's lawsuit against OpenAI. So in this, Sutska revealed that after Altman was fired in 2023, Anthropic expressed excitement about a potential merger that could have installed Anthropic's dario Amadei as OpenAI CEO. The talks, however, collapsed due to what he described as practical obstacles, which were likely investor complications from both companies massive funding rounds. And interestingly, Ilya's deposition also detailed deep mistrust within OpenAI's leadership. He including a 52 page memo he wrote accusing Altman of a quote, consistent pattern of lying and manipulating colleagues. Now, despite initialing Altman's initiating Altman's removal, Ilya later supported his reinstatement, which as we discussed, came after nearly all of their employees threatened to quit. So Paul, you had mentioned to me this was getting a lot of attention in AI circles online. Like what's getting people talking the most about this?
- [64:47]
A
I think just the fact that there's on the record testimony now about this. So when I first looked at it, there wasn't a lot that jumped out to me that wasn't in Karen Howe's book Empire of AI. So one, if you read Empire of AI that you probably get the gist of this. That being said, there was just some interesting context. So as you mentioned, he was a co founder and we've talked about Ilya a topic or two ago with Musk, Altman, Brockman and others. He then left to start Safe Superintelligence, which was valued at 32 billion in spring of 2025. A couple interesting notes. So Helen Toner, who was a board member when Sam was ousted, and Helen is someone I follow closely online, she did tweet for the record, for those dissecting Ilya's deposition, this part is false. And she was referring to her being the one that had sort of suggested this merger with Anthropic. She said, I wasn't the one who made the board Anthropic call happen. And I disagree with his recollection that board members other than him were supportive of a merger. My feeling at the time was that we were exploring crazy options because the situation was crazy. I had no intention of trying to merge the company with Anthropic. And as Ilya says in his deposition, the possibility was actively on the table for an extremely short time. Now keep in mind, this whole thing happened over like 72 hours. Like, Sam's fired, they go through all these different things and then Sam's back. She then said, I'm not planning to weigh in on details about November 2023 whenever they come up, but since this was sworn testimony and about my personal views, I wanted to make clear that it's incorrect. So then the part of the deposition I did read was Referring to this 52 page memo you mentioned, Mike. In essence, Ilya was asked by the independent board director. So at the time, there was three independent board directors, I think. And then it was Sam, Greg and Ilya, I believe, were the other board members. There were two or three independent board members. So the attorney said, all right, then you sent this 52 page memoir to the independent directors of the board, correct? He said, correct. He said, why didn't you send it to the entire board? Sutskov replies, because we were having the discussions with the independent directors, only the attorney, okay, why didn't you send it to Sam Altman Sutskova? Because I felt that had he become aware of these discussions, he would just find a way to make them disappear. He continued, so the way I wrote this document was to the context of the document is that the independent board members asked me to prepare it and I did. And I was pretty careful because at this point, Ilya had raised concerns, as had Mir Morati, that Sam might not be the right leader moving forward. Most of the screenshots that I have, most are all I don't remember. I get them from Mira Morati. So Mira was also sending screenshots to Ilya that was demonstrating Sam's inability to be effective leader. It made sense to include them in order to paint a picture from a large number of small pieces of evidence. The attorney. Okay. And which independent directors asked you to prepare your memo? Sutskova said it was most likely Adam d'. Angelo. Adam is the only remaining board member, by the way. So after all this hoopla, Adam remained in. In as an independent board member. The attorney said, all right. And the document that you prepared, the very first page says, quote, sam exhibits a consistent pattern of lying, undermines his execs and pitting his execs against one another. That was clearly your view at that time. Correct. And Sutska says, correct the attorney. And did you want them to take action over what you wrote, Sutsko? I wanted them to become aware of it. But my option or my opinion was that action was appropriate. And what action did you think was appropriate? Termination, said Tutskawa. Okay. You drafted a similar memo that was critical of Greg Brockman.
- [68:20]
B
Correct?
- [68:20]
A
Said yes. You sent that to the board. Yes. Does a version of your memo about Greg Brockman exist in any form? He said, yes. Somebody has it. I didn't know about the Greg Brockman one. That was the first time I heard about it. So my whole takeaway with this mike is, is if anybody watch the social network and the Facebook crazy story.
- [68:36]
B
Yes.
- [68:36]
A
Like we are. This is on steroids. Like the OpenAI movie is going to be insane.
- [68:42]
B
Yeah, no kidding. That just feels like an episode of Game of Thrones.
- [68:46]
A
Yeah, it's wild.
- [68:49]
B
All right, next up, according to Bloomberg, Apple is planning to pay about a billion dollars a year for an ultra powerful 1.2 trillion parameters artificial intelligence model developed by Google that would help run its long promised overhaul of the Siri voice assistant, according to people with knowledge of the matter. Bloomberg went on to say, quote, following an extensive evaluation period, the two companies are now finalizing an agreement that would give Apple access to Google's technology. So reportedly, Google's AI will handle Siri summarizer and planning planner functions, helping the assistant better understand context and execute complex tasks. Internally, this overhaul is codenamed Linwood, part of a broader project known as Glenwood, led by Vision Pro creator Mike Rockwell. And software chief Cred Federighi. And the redesigned Siri is slated to launch next spring as part of iOS 26.4. So interestingly, while Apple continues developing its own trillion parameter model for release next year, Google's AI will also apparently quietly run behind the scenes powering a smarter, more capable Siri. So, Paul, interesting, we're seeing some movement on this. I personally am just still a little confused. Like, they're developing their own model too. Like, I like, why are they even bothering given that they're already in this position, like, at some point. I mean, I'm not saying people should give up, but what are we doing here?
- [70:14]
A
Yeah, this is like Apple and Google have this really unique history where they, you know, do collaborate on things. This isn't news. Like, it became news again because someone else started talking about this. But Mark Gurman broke this in like April of this year that these, that the companies were working on something like this together. So he came out with an updated story after last week when everybody else was just running with this without crediting him with the fact that he'd already said this was going to happen. And I think we talked about the podcast like multiple times that this is the direction it was going. Yeah, I think they're just accepting that they're not a frontier lab. Like, they're, they're not going to build to compete with like what Gemini 3 is, because my guess is they're going to be building at least like a version of like what Gemini 3 is going to make possible. Yeah. And I think that Apple's going to make a bet on smaller models eventually are probably going to be sufficient. So, like whatever Gemini 3 or whatever version of Gemini they're going to get that'll run this thing, they're probably looking out 18 to 24 months and saying, well, we'll be able to run a 10x smaller model on device. It won't need to go to the cloud. Yeah. And so my guess is Apple's is going to focus in on that, like building these more efficient smaller models that can run on device, but in the interim, they need to fix Siri bad. And they've accepted that they just aren't going to get there on their own. And so I think it makes a ton of business sense. And again, they've done deals like this before on search and maps and other things. So I think it's good. I think Apple investors are just like, just fix it. Like, we don't care if it's your model or not. As a user of Apple, Apple products. Like, I don't care, Just make it work. Like, it doesn't matter to me what it's running on. Just make it functional. So, yeah, seems like the very logical choice from Apple.
- [71:57]
B
All right, our last topic today, we're seeing some stories where tech giants are working overtime to reframe the narrative around AI with a wave of positive announcements about jobs, education and economic impact. So first, Meta unveiled what it calls a $600 billion commitment to American infrastructure and jobs, touting its new AI optimized data centers as engines of growth. The company has published a specific report on this, saying its US projects have already supported 30,000 skilled trade jobs and added 15 gigawatts of power capacity and pledges to be water positive by 2030. So the message is that AI is not just about technological progress. They're tying it to American workers, communities and sustainability. Now, interestingly enough, on the same day, Google spotlighted a $5 million investment in Oklahoma to expand AI training and workforce programs through local partners. Now, both of these moves are happening amid heightened criticism that Big Tech's AI push threatens privacy, jobs, public trust, and as well as the environment. So we're kind of talking about this now, Paul, because we wanted to highlight the how they're actually starting to try to get ahead of possible societal backlash. Like you had mentioned this in our internal chat, in the case of Meta was kind of quote, them trying to get out ahead of societal revolt by positioning it as a job creator and responsible use of energy.
- [73:24]
A
Yeah, the first time I've actually heard the term water positive. So I don't know if that's like a common phrase, but I had to look it up. So a concept where an entity restores or adds more water to a watershed than it consumes or depletes, which, I mean, that's obviously that makes a lot of sense, but I'd never heard that talked about like that. Yeah. So I, I don't know. Like a pretty safe job for a while is going to be lobbying or working in community investment in these Frontier labs. Like, they are going to be working overtime on lobbying in Washington and they are going to pour money into community investment programs that they can invest into cities. Like what we saw Google do with Oklahoma. And PR staff for the frontier labs are going to be very busy. So. And again, I came from that world. I have, I haven't done lobbying, but I have done community investment. I led PR programs for some brands throughout our agency. I get how this all works. And I'm just telling you, if you don't live in this world or haven't lived in this world. This is how it starts. Yeah, you look at these very specifically timed, very crafted messages, very strategic investments in certain areas, specifically places where data centers are going to be built, where it's going to affect the environment. Like this is PR at its core. You try and affect the way people perceive things and not in a negative, I'm not saying this is a negative thing. This is just what PR people do. You try and affect, affect perceptions and behaviors and you do that in an environment like this where it's through lobbying and community investment programs. So I think we are going to see a flood of stories like these ones coming from the major AI companies.
- [75:08]
B
From a total realist perspective. No value judgment on any side of this. Like the reason this matters too is these are the talking points you are going to hear next year as we get into political, you know, whatever firestorm we get into over this. You're going to hear about jobs, water, positive, all this stuff. I bet you you're going to hear these terms again and again.
- [75:29]
A
Yeah, we are going to hear endless about how many jobs are created by data centers. Like that is like the most obvious talking about. We're already seeing it, how sustainable those jobs are versus like one time things to build the data center and they aren't needed after the 12 months, whatever. But yeah, so again, like a lot of times on this podcast we're just trying to zoom you out and, and show how this all is connected and you can see from the first topic of the impact on like OpenAI looking for that government support all the way back to this like lobbying and community investment. Like it's all interrelated and like I, you know, it's a fun part for me as a storyteller, Mike is like every week we do this podcast like almost it's kind of I think about mekon, like I think about the overall picture and the story and then like the topics become like this part of the story and something is not obvious, the connection between them. But I think the people that listen enough and like think about this the, the way we think about it each week you just start to like connect the dots and you see how this all is connected and it's a fascinating picture to be able to look at. It's not always like optimistic or it doesn't always make you like feel super excited about the near term future, but it's nice to at least see the picture when so many other people are just like, like, you know, oblivious to what's happening.
- [76:44]
B
Right. I couldn't agree more. That's my favorite part of this, is just the connecting the dots and getting to learn about so many different areas that AI touches.
- [76:51]
A
Yep.
- [76:52]
B
So to that end, Paul, thank you for connecting the dots for us this week. That's all we've got this week. Just a couple quick final announcements. If you have not yet subscribed to the Marketing AI Institute newsletter, it is called this Week in AI and we run down all the stories we talked about today, plus all the news we couldn't fit in the episode. So go to marketingai institute.com, and you get a nice weekly brief of all your AI news that you need to stay on top of these critical issues. I would also mention if you can, please give us a review on your podcasting platform of choice. It helps us get better, improve the show and get into the ears of more listeners over here. So if you have not taken a second to leave us a review, we'd greatly appreciate it. So Paul, appreciate you, appreciate you breaking down everything for us again this week.
- [77:40]
A
Yeah, and I'll add two more quick ones. Mike, again, the Pulse Survey. We'd love to have your feedback on the AI Pulse Survey. And then I am doing I do a free monthly Scaling AI webinar so we have a class coming up on Friday, November 14th. We'll drop the link in the show notes that is in partnership with Google Cloud, part of our AI literacy project. I teach a free intro to AI each month and a free scaling AI class each month. So November 14th at noon is a free It's a zoom webinar so it's easy to join. Again, we'll put the link in the Show Notes. But if you've got time on Friday and you want to go through five steps for scaling AI heading into 2026 planning, it's a great time to go through that class. All right, thanks Mike. We will be back with you all next week. 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 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. Until next time, stay curious and explore AI.