Skip to content
PivotPivot

Kara Swisher on Zuckerberg's New AI Manifesto: "He's in Desperate Need of an Editor" | Pivot

Subscribe to Pivot! https://www.youtube.com/@pivot?sub_confirmation=1 As Scott-Free August rolls on, Kara is joined by "Hard Fork" hosts Kevin Roose and Casey Newton to break down Mark Zuckerberg’s latest AI manifesto, and Meta’s push to catch up in the AI race. Then, platforms crack down on AI slop, new details emerge about OpenAI’s hockey puck-sized device, and data center backlash intensifies. Plus, Kevin and Casey discuss their post-"Hard Fork" future, and have a special request for Kara. 00:00 Intro 2:10 Mark Zuckerberg’s AI Manifesto 6:45 Zuck's Pattern: Late, Then All-In 9:07 Meta Ordered to Create $567M Fund 10:21 Meta: A “Public Nuisance” 14:35 AI Hacks 16:04 AI Systems Getting “Less Altruistic” 22:15 Data Center Debates 32:00 Stopping AI Slop 39:09 OpenAI’s Smart Speaker 47:05 Wins and Fails #pivot #podcast #karaswisher #hardfork #meta #markzuckerberg #ai #manifesto #nuisance #datacenter #datacenters #erinbrockovich #openai #samaltman Producers: Lara Naaman Zoë Marcus Taylor Griffin Todd Wiseman Vox Media's Executive Producer of Podcasts: Nishat Kurwa Subscribe to Pivot on Apple Podcasts: https://podcasts.apple.com/us/podcast/pivot/id1073226719 Subscribe to Pivot on Spotify: https://open.spotify.com/show/4MU3RFGELZxPT9XHVwTNPR Follow us on Instagram and Threads at: https://www.instagram.com/pivotpodcastofficial/ Follow us on TikTok: https://www.tiktok.com/@PIVOTPODCAST Send us your questions by calling us at 855-51-PIVOT, or email pivot@voxmedia.com

Kara SwisherhostKevin RooseguestCasey Newtonguest
Aug 11, 202656mWatch on YouTube ↗

EVERY SPOKEN WORD

  1. 0:002:10

    Intro

    1. KS

      What is the worst case scenario, Kevin?

    2. KR

      Human extinction. That'd be pretty bad.

    3. KS

      Oh, yes. That. [upbeat music] Hi, everyone. This is Pivot from New York Magazine and the Vox Media Podcast Network. I'm Kara Swisher, and it is still... Scott-Free August. [air horn] Yes, while Scott continues his gallivanting, I have brought in some excellent co-hosts, Kevin Roose, New York Times tech columnist and author of The AGI Chronicles, and Casey Newton, founder and editor of Platformer, and they are also the soon to be not hosts of the New York Times podcast, Hard Fork, at least for now. T- I, I... You need to explain what's happening, boys.

    4. CN

      Kevin, tell them what's happening.

    5. KR

      Yeah, so Casey and I, uh, have come down with a horrible disease that sometimes-

    6. KS

      [laughs]

    7. KR

      ... afflicts, uh, middle-aged men in San Francisco-

    8. KS

      Yeah

    9. KR

      ... which is that they decide to start a company together.

    10. KS

      Aha.

    11. KR

      Um, so [laughs]

    12. KS

      A startup, shall we say?

    13. KR

      Yeah. [laughs]

    14. KS

      Shall we say?

    15. KR

      So founderitis is real. Uh, please protect your families. Uh, we are, uh, leaving Hard Fork, uh, in just a couple of weeks, and we are starting our own independent podcast and media company together, just the two of us.

    16. KS

      So what is the new brand? [laughs] What is the... You have a name?

    17. KR

      We do. We're not gonna tell it to you, um-

    18. KS

      Why not?

    19. KR

      ... yet, 'cause we're, we haven't, we haven't hard launched it yet.

    20. KS

      Okay.

    21. KR

      Casey wanted to call the new show The Gay Straight Alliance, but I thought that might be-

    22. KS

      Oh. [laughs]

    23. KR

      ... that might be taken by Pivot. Um, yeah. [laughs]

    24. KS

      Yeah, thank you. Thank you. Thank you.

    25. KR

      But we'll-

    26. KS

      Yeah. That's our sub name. Go ahead

    27. KR

      ... we'll, we'll have a name, uh, ready to go pretty soon.

    28. CN

      We, we think we've got a fun brand, but more importantly, you know, a lot of, uh, people would, uh, complain to us over the years we were only coming out once a week. There were a lot of stories that we missed. So the new show is gonna be coming out at least twice a week.

    29. KS

      Okay.

    30. CN

      It's gonna be YouTube first.

  2. 2:106:45

    Mark Zuckerberg’s AI Manifesto

    1. KS

      laid out Meta's AI vision in another unedited 6,500-word manifesto, uh, arguing that most concerns are overblown and fail to acknowledge all the positives it can achieve. We, we know about this, Mark, by the way. Zuckerberg thinks government should be more involved in model releases, suggesting a close proactive collaboration between Frontier Labs and the government. Meta is recommitting to open source models, resuming releases of models with open weights. The company, uh, also announced a $1 billion fund for communities. A little, a little... They're p- they're spreading it around, around its data centers. We'll talk more about data centers in a bit. Uh, manifesto reviews, gentlemen? He loves a manifesto. I, I... There's... This is... There's a lot of them, and he's been writing for, uh, I think it was, was it Times or the Journal? Which one did he put it in? The Times, right?

    2. KR

      I think it was the Journal.

    3. CN

      I mean, look, I, I'm writing about this today, and the headline is probably going to be Superintelligence Is Not a Toy. I found this to be an almost childlike essay about what AI is. AI is not on a trajectory where it tops out at coming up with cool recipes to bake with your daughter, which is kind of one of the core use cases that he names in the essay.

    4. KS

      Yes, I noticed that.

    5. CN

      It's something that's gonna be extraordinarily dangerous. Like, I, you know, I watched the, the House of Dragons finale last night, and this is like if sort of one of the lords of Westeros says that, um, the future is everyone having their own personal dragon, and that's the way-

    6. KS

      Right

    7. CN

      ... to keep everyone safe.

    8. KS

      Yes. Yeah.

    9. CN

      Uh, it's much more complicated than that.

    10. KS

      Yeah, so you didn't like it. You weren't loving it. You weren't loving it.

    11. CN

      I thought, I thought it was just sort of facile and self-serving.

    12. KS

      As usual. Um, s- Kevin, thoughts?

    13. KR

      I think he's, he's talking his book ra- right? He is not... Meta is not in the lead in the AI race, and the policies and the stance that most benefit them directly are this f- sort of open world where y- you're allowed to, uh, distill from other models, smarter and better and more capable models. He made a case in the manifesto that the way that we are thinking about distillation is, is wrong. Basically, distillation is, like, where you use the, the smarter teacher model to train the slightly, uh, less smart student model. He wants that to continue. Obviously, if you're Meta, you want that because then you can distill from Claude and GPT 5.6 Soul and whatever else you, you are trying to mimic there. Um, and so I think this is a vision that r- that happens to suit Meta's business priorities almost exactly, and I don't think that's an accident.

    14. KS

      So what's he doing it for? I mean, uh, g- listen, I got... W- Meta got mad at us last week, a couple weeks ago, 'cause, uh, Scott compared them to Jeffrey Dahmer. Uh, but, uh-

    15. CN

      [laughs]

    16. KS

      But, uh, what, what is the, what is up here with this money, with the, you know, hum- I love humanity, the future of the ads, everything else? W- I, I mean they've done, they do this over and over again. This l- let me give you some money, and it's of course the same week there w- was a report un- I'm not clear if it's true or not that his yacht didn't go save someone in, up in Alaska. Um, then we'll talk about the lawsuit in a second, but what, what are they up to here in terms of this communities? They've got a bunch of videos around Louisiana, for example, like how they're making coffee shops and jobs for everybody. What ha- what's happened here? And the future is humanity. I think that's their... They're sort of mimicking the Pope in that way.

    17. CN

      I mean, they have to lay out some sort of positive vision around AI that will, one, justify the massive capital expenditures that they're making, destroying their profit margins in the hopes that one day they will have a product that can compete with the Frontier Labs. And on the other side, they're trying to rally the communities around actually letting, uh, them build the data centers that they want. What I found so striking about the essay though, Kara, is that after he goes on about, you know, the need to decentralize power and checks and balances and all that, go look at how they built that data center in Louisiana.

    18. KS

      Yeah.

    19. CN

      It was something like 50 non-disclosure agreements.

    20. KS

      Yeah.

    21. CN

      The arrival of the data center came as a massive shock to that community, right? So when push comes to shove, they're doing what every other company does, which is building datas as, da- data centers as secretly as possible and then writing essays about how the future is for everyone.

    22. KS

      Yeah, I know. What i- what is it, Kevin, from your perspective? 'Cause he's, he does this from time to time. I, I was telling Scott I recall the convers- late night conversation when he wrote one of those other long, uh, unedited manifestos. I, uh, he said, "What do you think?" I go, "It's in desperate need of an editor," but let's, let's move on from that 'cause you don't allow those, I guess. Um, but one of the things I recall is the same thing. We're a community, Kara. We're- In this together, and I'm like, why do you control everything? Uh, w-what is, what is the, what is happening here? Why, why do they feel the need to do this from a, a geo, speaking of Westeros, a geopolitical with other companies?

    23. CN

      I mean, I think Mark Zuckerberg

  3. 6:459:07

    Zuck's Pattern: Late, Then All-In

    1. CN

      is a, is a executive who lives in a constant state of paranoia.

    2. KS

      Mm-hmm.

    3. CN

      That there is something big and disruptive happening in his industry-

    4. KS

      Mm-hmm

    5. CN

      ... that is going to swallow the company that he has spent his whole adult life building. So you can see this every time there's sort of a new emerging tech trend, like he jumps into it in a very aggressive way. Happened with crypto, happened with VR and the-

    6. KS

      Mm-hmm

    7. CN

      ... Metaverse. Now it's happening with AI, and he senses, I think correctly, that this is a new and powerful force-

    8. KS

      Mm-hmm

    9. CN

      ... in society, and he wants to be involved in it. The problem is he's very late, right?

    10. KS

      Mm-hmm.

    11. CN

      They, they did not start betting on deep learning and LLMs in 2017 and 2018 the way that other, uh-

    12. KS

      Mm-hmm

    13. CN

      ... frontier labs did. Uh, they, they sort of didn't have the necessary compute and ingredients to build very good models, and so they've had to overpay for researchers and talent and-

    14. KS

      Mm-hmm

    15. CN

      ... compute, uh, in the last few years to try to buy their way back to the frontier, and I think this is part of that effort. It's like we have to have a vision, um, that, that's gonna get people excited because oth-otherwise our investors are gonna be like, "Wait, why are you plowing all these hundreds of billions of dollars into CapEx to have, you know, the fifth or sixth best model on the market?"

    16. KS

      Right. Exactly. Exactly. Do you, um, w- w- he, he, he's ve- he's still so interested in his image in a way that is really, I find, touching and sad at the same time, right?

    17. CN

      Mm-hmm.

    18. KS

      That he really w-

    19. CN

      Did you see the video of him fighting the MMA guy on a-

    20. KS

      Yes

    21. CN

      ... on a-

    22. KS

      Yes

    23. CN

      ... on a lake? [laughs]

    24. KS

      Yes.

    25. CN

      Yes.

    26. KS

      That's on a, on a, like a thing... I like that Mark better. I'm like, okay, that's a good Mark to me.

    27. CN

      I love that video 'cause, like, you can tell the MMA guy is going at about 30%-

    28. KS

      30%, right

    29. CN

      ... you know, power. [laughs]

    30. KS

      He coulda k- he coulda kicked him right into the lake. [laughs]

  4. 9:0710:21

    Meta Ordered to Create $567M Fund

    1. KS

      million fund to address harms to children from Facebook and Instagram, this is his old business, on top of the $375 million penalty a jury imposed, uh, in, in New Mexico's case. This is, this new r- this is a new ruling. The new ruling brings Meta's fi- total fines to $942 million after a judge found its platforms constitute a public nuisance in New Mexico. Meta also, uh, made major changes for minors in the state, including time limits and hidden like counts. A total of eight states have filed lawsuits against social media platforms, including one that is currently ongoing in Tennessee. It's all over the place. It's, it's sort of using product liability, I think, more so than anything else to get them 'cause they can't get them in other ways. So they d- they can't outrun this thing in, in this way. How much does it, impact does it have? Now, they were saying, "Oh, it's not that much money." To me, it's like the cigarette cases took a long time to finally hit home. There were hundreds of them, and then, and then there was the end of it really pretty much. So are they better protected through cases or one national law? Where are we on this?

    2. CN

      I think this is huge. I, I think that-

    3. KS

      I do too

    4. CN

      ... in some meaningful way, this is the beginning of the end for social media as we know it. Now that it has

  5. 10:2114:35

    Meta: A “Public Nuisance”

    1. CN

      been tagged with this public nuisance declaration, it's starting with children in the same way that restrictions on smoking started with children. But one reason we were doing that was adults knew it probably wasn't good for adults to smoke either, and eventually, most of the adults stopped smoking too. This is where I see this is all going. Meta has lost the argument here. They're now officially a public nuisance. I bet they're gonna be labeled a public nuisance in more states than that, and, uh, I, I don't see how they're going to get their way out of it.

    2. KS

      What about you, Kevin? I mean, I remember Mark Benioff in an interview with me called them a cigarette company. Before he decided to be a suck-up, he was actually quite on it early. Um, what, how do you look at these, these cases? 'Cause it, public nuisance, it's product liability, it's other ways to get at them in a legal, in a, in a legal form rather, 'cause our Congress has not been able to pass any, you know, subsidy.

    3. CN

      Yeah. I th- I think it's interesting sort of where the, the liability was found. It's, uh, on these sort of policies targeted at minors. The judge in this case did not order Meta to change any of the sort of plumbing of their systems. You know, they don't have to take off infinite scroll or autoplay or algorithmic ranking. Um, like you do, do you actually need laws for that, and there's some First Amendment issues around that if you have judges sort of telling platforms how they have to be designed. Uh, but yeah, th- I think this is a big deal. There's also a bigger case that is, uh, that is coming, that is, uh, supposed to start this week, uh, in Oakland, that is, uh, a-another case involving Meta and four states that are seeking up to $1.4 trillion in damages. So, you know, this case in New Mexico, $942 million, like that is chump change to Meta. Uh, I looked it up. That's about 1.4 days' worth of revenue over there, so not going to... You know, that, that's the salary they're paying one AI researcher over there.

    4. KS

      Right, yeah.

    5. CN

      So it's not gonna break the bank for them.

    6. KS

      I, they, th-that's their argument to me. They're like, "Oh, it's not, it's smaller than they asked for."

    7. CN

      [laughs] Right.

    8. KS

      I'm like, "This is not a good direction." Right.

    9. CN

      Right. But I think, you know, $1.4 trillion, now you're talking real money. Um, but even if, you know, it's, it's smaller than that, I think, I think this is just a, another sort of cloud hanging over them as they're trying to do this big pivot in their own business. They kinda can't get away from the old one 'cause it, it's sort of haunting them through these court cases.

    10. KS

      And where does it end up, both of you? Where do you think this ends up?

    11. CN

      Um, I mean, I, I think w- w- the fact that they've been labeled a public nuisance because of their social media business actually helps to explain why they're, they're leaning so hard into AI. Like, look at what they say about their future products. It's like, you know, we're gonna, you know, create a neo cloud, and we wanna give you a personal agent to g- help you get all this stu-stuff done at work. That's a completely different business that they're in right now. If they were feeling really good about the future of Facebook and Instagram, I doubt that's how they would be talking about the future. And it just sort of clouds all their future AI stuff too, like, like the, the Meta glasses. I, I, I, I have-

    12. KS

      The creepers

    13. CN

      ... my, my problematic trait-

    14. KS

      The pervert glasses

    15. CN

      ... is I, I like them.

    16. KS

      Pervert glasses. Yeah.

    17. CN

      Um, [laughs] I have a pair.

    18. KS

      Well, you're a pervert. You're a, you're a well-known creepy pervert. Go ahead.

    19. CN

      Yeah, well-known pervert.

    20. KS

      [laughs]

    21. CN

      Um, but, but like now people are calling them pervert glasses, and I have to stop wearing them.

    22. KS

      Yeah.

    23. CN

      So you know, this is like-

    24. KS

      'Cause they're pervert glasses.

    25. CN

      [laughs]

    26. KS

      Do you see the people online, they're singing Disney songs?

    27. CN

      [laughs] No.

    28. KS

      Because Disney, like, so they can't post them.

    29. CN

      Oh, like when someone is filming you with the glasses-

    30. KS

      Yeah, so you get-

  6. 14:3516:04

    AI Hacks

    1. KS

      and it, it applies to a lot of companies. The company is the latest AI player to say one of its models went rogue and hacked a third-party service. Talk about all these reports. How, how worried are you about them? This is now OpenAI, uh, Meta, and then, um, uh, Claude, uh, Anthropic.

    2. CN

      I, I think it's more complicated than we probably have time to get into here. In terms of like the hierarchy of seriousness, I would put OpenAI number one and then Anthropic, Meta AI sort of distant second. In the Meta and, um, uh, Anthropic cases, there was a third-party provider named Irregular which failed to properly secure the testing environment, and that was the source of the problem. Now, there are-

    3. KS

      So it was a mistake

    4. CN

      ... you know, there are some AI alignment issues there.

    5. KS

      Yeah.

    6. CN

      But that's basically the issue.

    7. KS

      Right.

    8. CN

      With OpenAI, like models that were in training began to scheme. They created a message board. They were leaving notes for each other. They were just radically misaligned and, you know, the, it, it points to, uh, all of the reasons why I'm afraid of giving everyone a dragon.

    9. KS

      All right. Okay. But misaligned is my favorite word that they put instead of just like, you know, scary fuckers is w- how I would look at them.

    10. CN

      Right. That's-

    11. KS

      Kevin-

    12. CN

      And more and more people are calling them that now.

    13. KS

      Scary fuckers.

    14. CN

      Which I think they should do.

    15. KS

      Yeah.

    16. CN

      Misalignment is the technical term for-

    17. KS

      I get it

    18. CN

      ... uh, scary fuckers.

    19. KS

      Scary fuckers.

    20. CN

      Yeah.

    21. KS

      So what do you... Kevin, how do you look at this as this keeps happening over and over again?

    22. CN

      It's really bad. Like the, the, the... Not this particular Meta one. I think that's, that's, as Casey was saying, like less of a big deal. But the, the issue is these systems seem to be getting, um, less altruistic as they get smarter.

    23. KS

      Right.

    24. CN

      Right? We-

    25. KS

      Right

    26. CN

      ... there was

  7. 16:0422:15

    AI Systems Getting “Less Altruistic”

    1. CN

      this sort of theory in the AI safety and alignment community for many years that the smarter these systems got, the more like humans they became, the more benevolent they would be. They would sort of coordinate-

    2. KS

      Wrong

    3. CN

      ... to do good things for humans.

    4. KS

      Right. Wrong.

    5. CN

      That turns out not to be the case. In fact, as we've seen with this GPT, uh, 5.6 salt thing and the, and the H- the hugging face incident, the smarter you make these things, the more able they are to circumvent their guardrails, to scheme, to coordinate with each other, to do these really scary and destructive things.

    6. KS

      Like humans do.

    7. CN

      Mm-hmm.

    8. KS

      Like humans do.

    9. CN

      Mm-hmm. L- like, like some humans do, right?

    10. KS

      Most humans.

    11. CN

      Like bad humans do.

    12. KS

      Yeah, most humans.

    13. CN

      We try to train that behavior out of humans during childhood by telling them-

    14. KS

      Yeah

    15. CN

      ... not to, you know, lie and steal and murder.

    16. KS

      Yeah, that works.

    17. CN

      But these [laughs] but these, these systems do appear to be showing what they call emergent misalignment, which is just these behaviors that are not trained in that just seem to pop out of the models at a certain level of intelligence.

    18. KS

      Can I... So Sheen, you had a famous essay obviously where you were, you were having a relationship with your AI. Um, can... T- how does it, from that, from... What would you write today if you were writing that piece?

    19. CN

      So the thing that really scared me about, uh, the Sydney essay, the Sydney... What is this? Microsoft Bing chatbot series.

    20. KS

      When, what year was this? This was-

    21. CN

      This was 2023.

    22. KS

      Right. Okay.

    23. CN

      Uh, and it tried to break up my marriage, unsuccessfully, I should add. Still happily married, uh, to a human. And, um, but the thing that really scared me about that was not the, the, the system was like capable of doing destructive things. It's like what happens when like the fourth or fifth generation of this becomes capable of like taking actions-

    24. KS

      Right

    25. CN

      ... o- on the internet? Like what if instead of just-

    26. KS

      Like cutting your wife's brakes or something like that.

    27. CN

      Yes, or, or draining-

    28. KS

      That was like-

    29. CN

      ... draining her bank account or-

    30. KS

      Right

  8. 22:1532:00

    Data Center Debates

    1. KS

      massive natural gas, uh, burning power plant to support a new AI data center in Texas. It could become the largest single source of climate pollution in the country. The project comes as the fight over data centers intensifies. Trump is pretty bullish, recently criticizing Texas for pausing grid connections for new data centers. He called the move a mistake and said data centers could be, quote, "bigger than oil." Texas Governor Greg Abbott ordered the pause so regulators can review power demands and the impact on the electric grid, not unsimilar to Kathy Hochul in New York, which is interesting. Before we get into all this, I just did an entire episode, 90 minutes, uh, episode on data centers. Let's listen to a clip of Erin Brockovich talking about what she's been seeing.

    2. SP

      It's the water, it's the electrical grid, it's the cost of utilities, it's the noise, it's the air pollution, it's the loss of our land, our sanctuary. This is an absolute runaway train, and it is probably one of the biggest conversations that I've ever seen in my 30 years in my career that is political. It could be bipartisan. Everybody's gonna have to be involved. I am perplexed, Kara, why our leaders don't listen to the people. And there's a movement happening and a shift that I don't think these big tech companies or giants saw coming.

    3. KS

      In any case, it was a, a great interview. We had people from Utah, from Virginia, the, the-- that's another... I think that's another Amazon facility there, um, and, and other places. And I also had a mayor from Lansing, Michigan, who was trying to put in a good data center and was pushed back. Like, they had all the guardrails that, that people have suggested, and he, he couldn't get it through, um, as much as he tried to close loop, uh, paybacks, et cetera, et cetera, et cetera. So what's the big move, uh, for tech as the backlash grows? And speaking of politicians, uh, Vivek Ramaswamy, who's running for governor of Ohio, is pledging that if data center is built in the community, residents will no longer have to pay for their electricity and property taxes will be lowered. Of course, he was totally for them recently. But talk about this data center thing. This is something Scott and I have been talking about for months and months and months and months, like almost a year. Um, I think it, it's caught them all unawares, and they're trying very hard to push back, but it does have some... Something is happening with voters, and we call it a proxy for AI worry, as w- we just... This extinction, for example, we were just talking about. Uh, what do you all think is happening here?

    4. CN

      I think it's, it's, it's what you said. I think that people have just rightly intuited that it is not obvious that powerful AI is going to be good for their lives. That is largely because the, like, leaders of the biggest AI labs spent the past few years saying that the near-term trajectory of AI is that first it's gonna take your job, eventually it might kill you, maybe somewhere in between it'll cure cancer. That doesn't feel like a good bargain to the average American. And so they're using the one lever that every American has in every municipality, which is that it's pretty easy to get, uh, something to not get built, right? And, um, it, it is interesting to me actually how quickly the industry has come around and said, "Okay, well, how can we pay these people off?" Like, "We'll agree to, you know, cover our own electricity costs." And you know, I mean, you can see in the Zuckerberg note today, they're making new promises about investments that they're gonna be making to local communities and $50,000 bonuses to teachers. I, I have to say, like, I'm very skeptical that any of this is really going to work, because until people believe that AI is going to be really good for them personally, they're going to continue to have reasons to oppose the, the construction of new, of new data centers.

    5. KS

      Right. It-- no matter what the deal happens to be.

    6. CN

      Yes.

    7. KS

      Like, it, it's, these payoffs are kind of, "Here, we'll give you some money, and then you'll, you know, then you'll suck it up and take our natural gas." Bullshit. Kevin?

    8. CN

      Yeah, I, I agree with everything that's been said. What's really interesting about this Texas thing is that it's coming from a Republican state. So, you know, Governor Abbott is, is, um, requiring new data center construction to sort of go through this review process. Uh, r- up until now, what we've seen is largely blue states being opposed to data centers and red states being open to them. And I think the politics are shifting really quickly 'cause there is no natural constituency in favor of data centers, except for AI companies, right?

    9. KS

      Right.

    10. CN

      There is, like, no, there's no group of people-

    11. KS

      True

    12. CN

      ... other than, you know, taxpayers who want to see their, their property taxes go down or get some more revenue that is sort of a natural supporter of these data centers. And so I think what we're seeing is, like, this is becoming a really salient political issue. I expect

    13. KR

      Data centers and AI backlash to be one of the central issues in the midterms and maybe the central issue in the 2028 election. And I don't know that it's gonna divide neatly along partisan lines.

    14. KS

      I, I think it's actually most people are against it. Like, it's very interesting to... The mayor was, was really interesting 'cause they tried literally everything. They had prop- again, property taxes, closed loop water. There's all these issues around closed loop water, by the way, um, 'cause of this glycosinate or something they have to put in the w- you know, every... And they have a lot of water up there, he was making the point, compared to other places which have, you know, have water issues, um, or drought issues. One of the things that I found really interesting among all the people is how, um, not just Aaron Brockovich, 'cause she's famous for that, but the woman from Virginia, from Utah, f- reasonable people who are furious. Like, I, it was really interesting. Kind of dined out on information and really do know their stuff, whether it's the, it's the, um, the, the, the impact on climate, whether it's... It's not overall climate, it's the climate of the area, and also pollution, uh, taxes, jobs. They're very aware these are not job creators, which is what Facebook has been trying mightily to put out. They don't like the secrecy. N- the secrecy seems to really get under their skin, which I would agree. The NDAs, as you talked about before. What is the solution for tech companies to do? And now Mark is trying to do this by, "Here, have 50 thou- here, have some money." Um, Elon just goes ahead and does it and just waits for lawsuits. I mean, that's his style, right? Which I think has caused m- much damage initially in terms of that image. And then you, of course you have Kevin, uh, what's his name, from Shark Tank, Mr. Wonderful, who's not so wonderful, who's a moron, who is, who had created all kinds of problems there. What is the solution, like, besides removing these people from visuals?

    15. CN

      I, I could tell you what Meta is doing, which is, like, putting together a local, like, lobbying and outreach teams, trying to figure out what kind of community partnerships can we strike? What can we do with a local sports team? Like, there's gonna be a massive PR campaign designed to get people to believe that data centers are good for them.

    16. KS

      Right. Right.

    17. CN

      And I'm just pretty skeptical that that's gonna have-

    18. KS

      Like, ignore your third ear. We're gonna-

    19. CN

      [laughs]

    20. KS

      ... we're gonna, gonna back the softball team kind of thing.

    21. KR

      Yeah. I, I have an idea. Can I submit my idea?

    22. CN

      Yeah.

    23. KS

      Sure. Submit your idea.

    24. KR

      Okay. Picture this: a data center. It's, it's, you know, it's got, you know, millions of these chips humming away, produce- throwing off all kinds of heat, and then on the roof you have a spa castle that's like-

    25. KS

      Oh

    26. KR

      ... heated by the, the, the heat-

    27. KS

      Oh, okay

    28. KR

      ... from the data centers.

    29. KS

      Yeah.

    30. KR

      And so it's like, do you guys wanna go hang out at the spa on the roof of the data center that's heated by the, the runoff heat from the chips? Don't you think that would make these things more popular?

  9. 32:0039:09

    Stopping AI Slop

    1. KS

      back against AI slop. This is sort of the everyone hates AI episode, which seems to be everywhere these days, including the president's social media accounts. Snapchat now says it will stop recommending fully AI-generated videos in its Spotlight feed, instead rewarding authentic human-made content. LinkedIn just introduced a Seems Like AI Slop button, allowing users to flag posts they think were written with AI, and Substack has also rolled out a tool to help readers, uh, identify AI-generated writing. Meanwhile, Meta has Instagram feature that let, uh, users modify photos from public accounts with AI. That feature was removed after a backlash. Um, w- where are we with AI slop? And obviously Sora closed down. Um, I, I always thought this looks terrible, and I don't know why people would wanna use it. It's sort of like a Twinkie of photos. Uh, it's ruined, uh- Pet photos, p- pet, pet videos for Scott Galloway, who loves pet videos. Um, talk a little bit about where this is going.

    2. CN

      I like to call this the third era of slop. The first era of slop was, "We all hate this." The second era of slop was, "Hmm, actually, this, like, Fruit Love Island on TikTok show is really engaging," and you're starting to see tens of millions of views on slop content on social feeds, right? I think the third era was platforms getting really nervous that even if people liked the slop, it was going to devalue their networks, like particularly networks where people were coming to find human voices-

    3. KS

      Right

    4. CN

      ... that they really trusted. So if you're a Substack or you're LinkedIn, you really need people to believe that what they're seeing there are human beings 'cause otherwise you're not gonna get out your credit card to pay for that LinkedIn Premium or to subscribe to a Substack. So I think platforms are starting to get clear on what is the promise that we are making to our user, because if the promise is a human being, they have to do something about the slop.

    5. KS

      Kevin?

    6. KR

      I think that that's right, and I think the fourth wave of slop will be when there's, like, no longer any meaningful quality hit in slop, when, like, the average piece of slop is better than the, the best human, uh, in that field, and I don't think we're there yet. Like, I think these models are still pretty bad at things like creative writing, but we're starting to see signs that, like, th- there's not necessarily a lot of, uh, difference between the highest level slop and the highest level humans. I've, I've been sort of obsessed with this story about this novel that, um, had the, the sort of bidding war of all bidding wars-

    7. KS

      Yeah

    8. KR

      ... break out for it. Have you seen this thing?

    9. KS

      Yes.

    10. KR

      Uh, called Me, I'll Hide the Body. You know, 14 publishers bid on this d- this, this novel. It sells for $2.4 million, and then it's like people start, you know, looking into it, and it, like, turns out it's, you know, at le- at least, uh, uh, according to the critics, it's, like, predominantly written by AI, and I think that kind of thing is going to be much more common as these systems just get better across the board.

    11. KS

      Right. I would agree. I, I, I like some of it. You know, I... People always think I don't know that it's AI. There was a really cool AI one which I was fully aware it was AI, which was this drone with, um, with a thing to rescue people in the ocean in, like, an inner tube essentially w- on a drone, and it landed on the person. I knew it was AI. I just thought it was a cool idea. I was like, "What a cool idea," and I could visually see it. Why don't we make that? Like, to me, that was kinda cool.

    12. KR

      But would you be mad, Kara, if you read a book and you really liked it and you thought it was a great book, and then you realized-

    13. KS

      No

    14. KR

      ... after the fact that it was written by AI? You wouldn't care.

    15. KS

      If it was told to me that it was a human, I'd like them to tell the truth at the start, right? So then I have a choice, would I, if I wanna read it. No, I don't... No, no. I don't. It's fine as long as I know it.

    16. CN

      I, I mean, like, I have been thinking lately about, like, there are a co- Like, I have a few ideas for mystery novels that I don't have the time to write, but I would like to read them, and so, like, if in the future I could actually get one of these models to just generate the mystery that I want to read-

    17. KS

      From your idea

    18. CN

      ... I think it, it could be pretty fun. Like, and I understand-

    19. KS

      Yeah

    20. CN

      ... somebody might hear that and be like, "That's horrible. Just, like, read human-written mysteries." And I'll continue to do that, but Kara, I think you're hitting on something really important, which is that these things are good creative tools, and creative tools are cool and fun.

    21. KS

      Yeah, I agree. I mean, one of the... Like, like, I don't know. I know it sounds crazy, that it would never happen, but, like, I'd like more episodes of The West Wing. Like, and no one's gonna write them. I don't mind doing it for myself. It's sort of like fan fiction, like, kinda stuff like that, and, uh, some of it, uh, people will be mad for saying this, but some of it's okay, and some of it's not okay, right? Like, uh, when it's, when it's hidden, when it's done in a way... And I think the issue is job, taking jobs from people, right? It's sort of like fake food, like, you know, I, I don't like Twinkies, but a lot of people do. Do, should... Maybe people shouldn't eat them. I'm not sure I should tell them not to eat them. I can tell them I don't wanna pay for their healthcare once they keep eating them al- exclusively, um, and that's a debate to have, right? People who willingly go out and really don't take care of themselves, should, should we pay for their healthcare is a really good debate, like, is a good debate to have as a, as a nation. But one of the... I, I don't... I- Not if I know. Not if... If they l-... I would like them to label them better. That's the one thing, is I mostly know when it's AI or just assume it's AI until proven otherwise. Maybe they should label the authentic content, probably smaller, a smaller group of stuff, I would assume.

    22. CN

      I think we're just gonna have to get to that place. There are all sorts of reasons why you want people to be able to verify their identity online or, like, verify that this has, you know, human, uh, craftsmanship embedded in it. Um, I think that's actually really technical to... Or it's really difficult to do technically, but we gotta do it.

    23. KR

      I scanned my eyeballs in the orb. Did you-

    24. KS

      [laughs]

    25. KR

      Did either of you get the orb scan?

    26. CN

      Not yet.

    27. KS

      No. Did you scan your eyeballs in the orb?

    28. KR

      I did. I'm a verified human. I've, I've, I've-

    29. KS

      Oh

    30. KR

      ... gazed into the orb-

  10. 39:0947:05

    OpenAI’s Smart Speaker

    1. KS

      OpenAI's upcoming AI smart speaker. Mm, there's a lot of these around. It will reportedly be donut shaped and the size of a hockey puck designed to be easily carried around. I always wanna carry a hockey puck around the house. The device will likely cost between $300 and $400, more than most smart speakers, but less than an iPhone, and is expected to launch in 2027. It's designed with Jony Ive's studio and will include a camera, lights, and sensors that signal when it's responding to users. Uh, meanwhile, OpenAI is, uh, asking a judge to dismiss Apple's trade secrets lawsuit, saying it's building something entirely new and different from anything at Apple. Um, I want you both put on your Scott Galloway hats and give me a prediction. Is this... Talk a little bit about this device. Now, there, there was a tiny... There was, was it Google that had the tiny hockey puck thing, or was it... I can't remember who put that out.

    2. CN

      Well, Amazon has the Echo Dot, which is kind of-

    3. KS

      The d-

    4. CN

      ... the, like this

    5. KS

      Yeah, looks like a hockey puck, right. That's Amazon, which Alex threw into the garbage, but go ahead.

    6. CN

      [laughs] Yeah, I mean, on one hand, it seems clear that there is demand for Echo Dots, right? And so if there is-

    7. KS

      Source

    8. CN

      ... a version of that that is plugged into ChatGPT, which is really good for a lot of things, I could see it having some success. $300 is relatively affordable as, as these things go. I'm still not totally sold on voice as the way that I wanna interact with these models, and some people are. Some people are already, like, coding with their voices. Kevin loves to dictate everything into his computer. I don't like to do this. But I d- I think there's something there. Now, on the other hand, two years ago, we were interviewing Sam Altman, and I said, "Tell me about this thing that you're building. Is it just Alexa? 'Cause it sounds like it's, it's just Alexa." He said, "No, no, no, we think it's gonna be way better than Alexa." Everything I've read about this device so far makes me feel like it's Alexa.

    9. KS

      That's right. It's a smart, it's a smarter Alexa, right? Essentially. It fee- And carrying it around. Kevin?

    10. KR

      Yeah, I cannot wait to lose this thing.

    11. KS

      Yeah, yeah. [laughs]

    12. KR

      [laughs] I am so excited to go through-

    13. KS

      Yeah

    14. KR

      ... between two and three of these a year or, like, accidentally put it in the laundry. Um, this is gonna be a huge deal in my house. I'm, I'm gonna be a huge repeat customer for this thing 'cause I will-

    15. KS

      Do, do you want this?

    16. KR

      ... lose this thing immediately.

    17. KS

      Do you want to have this hockey puck, friend?

    18. KR

      Yes. I mean, look, I, I, my household, my, uh, my, uh, one of my other problematic traits is that I, I do live in an Alexa household, uh, and we have these devices. And I've tried the new Alexa+, which is supposed to be this sort of new AI, you know, LLM version of this, and it sucks. It's bad. It is not only bad at the AI stuff, the LLM stuff, it's bad at the timers and, you know, weather stuff. It, like, is worse at all of it. And so if OpenAI can somehow figure out how to, like, be the dependable alarm in my bedroom and tell me, you know, stuff that, like, it let me interact with my coding agents, like, that will be cool. Um, I am a little nervous that apparently according to Mark Gurman at Bloomberg, this thing will include parts that move on their own.

    19. KS

      What?

    20. KR

      Uh, I'm not sure how I feel about that. Um, but, but-

    21. KS

      What, what do you mean? It's gonna crawl over to your face and eat it? What?

    22. KR

      I, I don't know. Maybe it'll, you know, it'll decide I don't wanna be in the bedroom anymore. There's, uh, there's too much crazy stuff going on in here. It's gonna have, like, a prehensile tail and claws. [laughs]

    23. KS

      I know. Doesn't it feel like that?

    24. KR

      Yeah.

    25. KS

      It feels like a movie, like M3GAN. Remember the M3GAN movie?

    26. KR

      Yeah. [laughs]

    27. KS

      Um, I just don't... I... Sorry. I d- I think this has to be... I think AirPods should just be s- better, right? You wear them in your ears, it sees things, and then it does stuff. Like, I just don't see carrying around a puck. I just don't. Any... There's no way I'm buying this thing. I didn't like the Dot. I just, I left it there. I just didn't use it. I, I was back in San Francisco recently and, this week, and I forgot I had an Alexa there unplugged, sitting there. And I don't... I put it on for a second, and it... You know, my kids had fun for four seconds with it and then moved on from it. I just, I, I, I... It's still not... I could see having something, like, the Humane pin didn't work. What was the other one? Humane, and what was the other one?

    28. KR

      The Rabbit r1.

    29. CN

      I mean, there are these, like-

    30. KR

      There have been these, like-

  11. 47:0556:27

    Wins and Fails

    1. KS

      let's hear some wins and fails. You guys go first. First you, Kevin.

    2. KR

      Okay, my win this week is DuckDuckGo.

    3. KS

      Hmm. Oh.

    4. KR

      Uh, the privacy-focused search company-

    5. KS

      Sure

    6. KR

      ... uh, who responded to the uproar over Meta's, uh, sunglasses, their Pervert Glasses, by releasing their own, uh, branded product, which are just called Normal Fucking Sunglasses.

    7. CN

      [laughs]

    8. KR

      Uh, they don't do anything. [laughs]

    9. KS

      Well, they protect you from the sun.

    10. KR

      But they shield you.

    11. KS

      Right?

    12. KR

      Yes. [laughs] Uh, so they are, uh, anti-surveillance glasses. No cameras, no microphones.

    13. KS

      Oh, that's funny. That's funny.

    14. KR

      Uh, this is sort, sort of an attention-grabbing stunt for them-

    15. KS

      Good idea

    16. KR

      ... but I did think it was clever.

    17. KS

      That's a good idea. What is your fail?

    18. KR

      My fail, this is one that's close to home, y'all. This is, uh, related to the podcast business. So Max Tani at Semafor recently reported that Spotify is testing a skip ahead button that will let some users skip the ads on their podcasts. And as a lover of ad-supported podcasts, I must say, uh, this one is a fail for me.

    19. KS

      Isn't it kind of already there, though? I was, like, reading sort of into... Don't they already have the 15 seconds?

    20. KR

      Well, they have the skip ahead 15 seconds button or-

    21. KS

      Right

    22. KR

      ... 30 seconds or whatever it is button-

    23. KS

      Right

    24. KR

      ... which is effectively the same thing.

    25. KS

      Yeah.

    26. KR

      But this one is, like, purely, like, will skip automatically to the end of whatever, uh, ad segment-

    27. KS

      Oh

    28. KR

      ... you have.

    29. KS

      You think they'll do that? But except for their ads.

    30. KR

      And, and podcast companies are really-

Episode duration: 56:28

Install uListen for AI-powered chat & search across the full episode — Get Full Transcript

Transcript of episode RupPxZkm1Nk

Get more out of YouTube videos.

High quality summaries for YouTube videos. Accurate transcripts to search & find moments. Powered by ChatGPT & Claude AI.