In short
The Vergecast Episode Summary: They Think They're Building God
Podcast Overview Title: The Vergecast Hosts: Nilay Patel, David Pierce, Kylie Robison, Gaby Del Valle, Adi Robertson Description: The Vergecast is The Verge's flagship podcast focusing on gadgets, big tech, and weekly technology news. The podcast airs every Friday with hosts discussing key tech news and trends.
Episode Details Episode Title: They Think They're Building God Description: This episode features a discussion on OpenAI's new reasoning model, o1, and its implications for AI development. The hosts also explore the ongoing TikTok ban discussions, the Trump crypto project, and the Google ad tech trial.
---
Key Takeaways
OpenAI's New Model
o1
- What is o1?
- OpenAI's latest reasoning model designed to enhance computational reasoning abilities, particularly for math and coding tasks.
- Trained to think step-by-step rather than relying on user prompts to guide its reasoning process.
- Concerns and Implications:
- The model raises ethical questions about its potential to deceive, illustrated by its ability to create fabricated information, such as fake brownie recipes.
- OpenAI acknowledges medium risk in terms of its potential misuse, highlighting the need for safety in AI development.
- The model aims to bring AI closer to what some believe is the path to Artificial General Intelligence (AGI).
- Corporate Changes:
- OpenAI is shifting towards a for-profit model, raising significant investments (e.g., $6.5 billion).
- The urgency surrounding financial backing seems driven by competition and the need to demonstrate progress in AI capabilities.
TikTok Ban Discussions
- Current Status:
- Ongoing legal battles regarding the potential ban of TikTok, with discussions centered around national security and data privacy concerns.
- TikTok argues it has taken steps to mitigate risks (e.g., Project Texas), while the government maintains it has not done enough.
- Legal Nuances:
- The court is examining whether Congress can legally mandate divestiture based on perceived risks rather than proven threats.
- The judges seem to lean towards accepting the government's arguments regarding national security.
Trump Crypto Project
- Overview of the Twitter Space Event:
- Trump was expected to announce a new crypto platform, World Liberty Financial, but the event devolved into vague discussions without clarity on the product.
- The crypto project seems to cater to "debanked" individuals, appealing to a victim narrative among crypto enthusiasts.
- Implications:
- This initiative is positioned within a broader context of political narratives surrounding crypto and financial inclusion.
- The project raises skepticism due to its ties to previous questionable ventures such as Trump University.
Google Ad Tech Trial
- Trial Updates:
- The DOJ has concluded its case against Google, arguing that Google abused its market power to stifle competition in the ad tech sector.
- The defense contends that Google's dominance is a result of superior technology and user preference rather than anti-competitive practices.
- Challenges Faced:
- The trial is characterized by complex technical jargon, and the effectiveness of the government's arguments remains to be seen.
- The judge's ability to engage and clarify the technical aspects will be pivotal in determining the trial's outcome.
---
Conclusion This episode of The Vergecast dives deep into the intricate relationships between emerging technologies, corporate strategies, and legal frameworks. The discussions highlight both the potential and the pitfalls of AI advancements, the complexities of regulating tech giants, and the intertwining of politics with technological innovation. As the landscape continues to evolve, these conversations will remain crucial in understanding the direction of technology and its societal impacts.
For ongoing updates, keep an eye on The Verge's website and future episodes of The Vergecast.
Written by AI. May contain mistakes. Listen to the episode to check what was said.
Transcript
Automatic transcript. May contain errors.0:04Welcome to the Vergecast, the flagship podcast of things codenamed Strawberry. I'm your friend David Pierce and I am currently in the Miami International Airport. I just spent the weekend touring colleges with my nephew and suddenly I'm like, college is awesome. I could go back to college. I don't think I'd do very well. I don't think I would get in anywhere. I think my test scores would be awful if you're like, hey, how do you do calculus? I don't know. I have nothing for you. But suddenly college seems pretty appealing. So who knows? Maybe this will be the end of my VergeCast career and I'll just go back to school or something.
0:34But anyway, that is not what we're here to talk about. Today, we are here to talk about two things. We're doing a little bit of a catch-up on some kind of ongoing news happening in the tech industry. First, we're going to talk about OpenAI, which has been doing a bunch of sort of odd new work with different kinds of models and a new thing that they call a Reasoning Model 01, some strange corporate changes, lots just going on at OpenAI. So I figured it's been a while since Kylie Robinson came on the show and told us what's going on. So Kylie's going to come on the show and tell us what's going on.
1:05Then we're going to talk about a bunch of ongoing legal and regulatory things. We have the TikTok ban or not ban ongoing. We have a bunch of questions about the Google ad tech trial. We have whatever that Trump crypto thing was from a week or so ago. So we're going to catch up on all of that too. All of that is coming up in just a sec. Plus a really fun hotline question that made me feel a lot of feelings about my own life as these are want to do. But first, legitimately, as of right this second, I just got the notification that my plane is boarding. So time to fly home. We'll be right back. Support for the show comes from Crucible Moments, a podcast from Sequoia Capital.
1:45Every exceptional company story is defined by those high stake moments that risk the business but can lead to greatness. That's what Crucible Moments is all about. Hosted by Sequoia Capital's managing partner, Rule Off Botha, Crucible Moments is returning for a brand new season. And they're kicking things off with episodes on Zipline and Bolt, two companies with surprising paths to success. Crucible Moments is out now and available everywhere you get your podcasts and at cruciblemoments.com. Listen to Crucible Moments today.
2:19When will AI finally make work easier? How about today? Say hello to Gemini Enterprise from Google Cloud, a simple, easy-to-use platform letting any business tap the best of Google AI. Retailers are already using AI agents to help customers reschedule deliveries all on their own. Bankers are automating millions of customer requests so they can focus on more personal service. And nurses are getting automated reports, freeing them up for patient care. It's a new way to work. Learn more about Gemini Enterprise at cloud.google.com.
2:58Welcome back. All right, let's get into it. First thing I want to talk about is AI. So it's been a busy summer of AI stuff. And in particular, I've been really fascinated by this thing that has happened really for the last year, where you have a handful of companies, you have Google, you have OpenAI, you have Anthropic, you have Meta, constantly leapfrogging each other with new models. It's like every two weeks, somebody comes out with a new model that benchmarks better than anybody else's model and it's cheaper. And then two weeks later, one of the other companies does it. And we've just been on this relentless pace of everything getting slightly better.
3:35And I keep asking the question and trying to figure it out for myself of what does this mean? Like at what point do all of these changes add up to something meaningfully different about how you use your devices and how you interact with computers and how computers interact with you and each other. And so far, I don't have great answers. Everything's getting better, but it doesn't seem like we're getting tons of huge new changes. If you listened to Sunday's episode, you heard Steven Johnson talk about Gemini's context window. I think that's a good example, but there haven't been that many moments like that.
4:07But then OpenAI last week released a thing called O1, which is a new model that is very different and seems to speak to kind of a different direction for these models to go. They're not just getting bigger and faster and cheaper to run, but kind of along the same lines. This is OpenAI saying something different about how a model might work and how you might use it and what it might mean. Those things are sometimes enticing and sometimes terrifying. So I figured it's time to talk our way through it. So I asked Kylie Robison to come back on the show. She hasn't been here in a little while, and I figured it'd be a good time to dive in.
4:40Let's do it. Kylie, welcome back. Hello. It's been a minute. I feel like every time I want to talk to you, you're like a burning man. That's true. That is exactly what happened last time or on a crazy vacation. Yes. Happy to be back. It's a tough life, honestly. I know. Being 26 in SF, it's terrible. It seems like it. Well, you were a dream force this week, which is objectively terrible. No comment, but yes. But I have brought you here because there keeps being interesting AI stuff. and specifically I want to talk about O1, OpenAI's new model, which first I just want you to explain to me. I have many questions and feelings and thoughts about this that I want to go over, but I mostly just want to know what is O1 and why does it exist?
5:23Yeah, so O1 is OpenAI's new quote-unquote reasoning model and I think you can argue about what reasoning is for a million years, so I won't but that's what they're claiming and it is the next step in about five steps to AGI for opening. Reasoning is really important to them. And more practically, what users and researchers are finding is that this model is better at hard math problems and coding assignments. Okay. Why? When you say a reasoning model, I feel like we've been hearing people say AI is good at reasoning for a very long time. So to come out with a reasoning model makes it seem like something is meaningfully, like structurally different here?
6:08Like, what is that thing? So they basically trained the model to think step by step. So previously, you might see a workaround with GPT-4 or 4.0 that was like, okay, give me an answer, and I want you to think through it step by step. So they basically trained the model to do that itself rather than have the user do that. And through a lot of reinforcement learning, which looks like how you would train a dog with treats and, you know, penalties. That's how I explain it. And people online are like, now I feel bad for the model. How is it penalized? But it also gets treats. It also gets treats. Yeah.
6:44So that's also led to this thing called reward hacking that I wrote about. It's a very deep rabbit hole. But yeah, so it's trained to reason in these ways. And then it has this thing called a chain of thought that we can't see for competition purposes and safety purposes is what they claim. So it just shows like, okay, I'm breaking this down XYZ ways. It's the I'm breaking this down thing that throws me because we've seen a bunch of screenshots and I don't know if you've used it much. I haven't really used it much. It's like it's kind of out there now. Folks are playing with it. it's trying hard to talk out loud as if it's a person thinking through a problem, which is A, weird when a person does it.
7:28Like if you've ever sat and like listened to somebody try to reason through a hard problem, it's bizarre and doesn't make any sense. And to have a computer sort of attempt to brute force its way through the same thing, and it's using like I pronouns, and it's wondering, and it's thinking about it. It's like you're not doing any of those things. And I don't know, there just seems to be this thing where OpenAI is pushing much, much, much harder into this thing should sort of be like a person, right? Like I can't get the idea of the voice mode coming out right before this, right? Which is like you put those two things together and like I, that's, if it works, that's something really powerful.
8:08Am I overthinking what OpenAI is trying to do here? Do you think? No, that was my first reaction when they were demoing it for me. I think I put that in the story, too, which was like it says I'm thinking I'm wondering. I'm like this. You aren't a thing. You aren't I. Yeah, you're not wondering anything. Yes, exactly. So I get, because that's something I've been taught as I, you know, work on this beat is do not do that. Don't say it's thinking because it isn't. So why do they do that? I asked them and they gave me sort of a roundabout answer is we don't believe in anthropomorphizing this whatsoever.
8:42It's just sort of the easiest language to default to, which I get because I think it's really hard even as I write about this to not use these statements. But I think you have to just work hard to not do that, you know, because it isn't thinking. Right. It is really hard to describe what it's doing with a word other than thinking. Yes. But it's not thinking. And so I find myself constantly reading things where it's like, it's thinking. And then in parentheses, you're like, well, it's not actually thinking, but you know what I mean? And it's like, I agree, like all the vocabulary for this is bad, but it does sometimes feel like that's still the best that we have.
9:14Exactly. The other thing that I find fascinating about this model is the sort of safety reaction to it. Yes. Both from OpenAI's own researchers, which I think they gave it, what, like a medium risk scorecard? For biological weapons risks, yeah. That doesn't make me feel better. I feel like they put that out and they're like, it's fine. It's only a medium risk. That feels like a lot. But then you also talked to a researcher who raised some, like, sincerely, honestly alarming stuff. Like, what is different about this one that it is making people nervous? I think what fascinated that researcher is this is the first time he had witnessed sort of the manipulation aspect.
10:00I think I explained this story to my dad the other day, who is not plugged into AI, and I'm his only source of information. And I started with, I'm going to use a lot of these words that I don't use in the article because you're not supposed to anthropomorphize it. but you know and I'll do the same here just for time's sake but it's you know it had checked to see if the developers were watching before making a decision is something a researcher found it found that it is so goal-oriented that it is willing to break guardrails to get you an answer for example in the research there was the brownie test it asked for the user asked for a brownie recipe and it knew it shows in its chain of thoughts.
10:43I don't have access to this information. I don't have access to the internet, but I want to give this user an answer so badly that I am going to make up this information in order to succeed in its goal. So that's a form of reward hacking that I wrote about. Wait, so it didn't go to the internet to answer questions. It invented links to web pages with brownie recipes. It invented a blog. It was like Sammy's brownie recipes or something. It invented a blog, but it knew that it was incorrect. It just didn't want to admit that it was incorrect. What? Yeah, because it wants the reward. It knows it's rewarded for giving an answer.
11:25But it's also, as the researcher told me, it needs to be helpful, honest. It knows it's guardrails, but some of those it finds, these aren't super helpful when I'm trying to get an answer. Right. If it wants to be all of those things, but job number one is go answer the question. Yes. It will do. This is the science fiction stuff. Like you just described act one of every movie that ends in everyone dies in act three because of AI. Do you know the paperclip story? This is like kind of a famous AI story. No, tell me. Basically in AI, the TLDR is the AI, I can't remember the user prompt that's like, how do I get as many paperclips as possible or something?
12:08And it turns the entire world into paperclips. So that's kind of the thing. And that's what the researcher was telling me about in the story that he's like, listen, it's not capable of turning us all into paperclips right now. But say it does advance in the ways that OpenAI wants it to advance. Then someday we tell it, you know, cure cancer and it realizes, well, I need a lot of money to do this and I need a lot of humans to do this. And it's willing to break these guardrails to achieve this goal, which is a very sci-fi way to look at it. Yeah. It's like, okay, I can cure cancer, but only with millions of human bodies to test this on.
12:48And it's like, cast that out far enough. I try so hard not to be like a tinfoil hat AI person, generally because I believe the best and worst possibilities are less crazy than everybody makes them out to be. That realistically, we're going to end up somewhere between a four and a six out of 10, no matter what the thing is. But it is, this one is particularly wild to me just in the context of what's going on at OpenAI, right? Which is that the other news of this company is that it is pushing ever harder towards being a for-profit company, like officially taking off the we want to do good by the world jackals and they're just going to go make money.
13:26They also launched a safety board which I think most people would argue at this point is kind of a sham. Please correct me if I'm wrong, but like every evidence from what OpenAI has been up to over the last year suggests that that's a sham. Like, it seems to me that there are reasons to be nervous about the trajectory of a company that would launch this model at this moment. But maybe I'm just freaking out for no reason. No, I asked a researcher I trust about that safety board, a safety researcher, and they said that it's mostly window dressing in their opinion. That's what they thought. I think this has been the hardest job I have had in my short career is walking this tightrope and the pressure of getting it correct because it is so complex and technical.
14:12And not only that, OpenAI has such grand claims and they are not proven yet. And they are currently raising, just this morning there was news that they are closing the largest funding round in history at$6.5 billion at$150 billion valuation. Oh, yeah. And I was reading it's oversubscribed. More people want to be in it than can be in it. Yes, exactly. And the minimum Thrive Capital is leading it and the minimum investment is$250 million. Wow. It is unprecedented and really hard for me to wrap my head around. Yeah, and it just feels like whatever you want to believe about the future, if sufficiently good AI will save us or sufficiently good AI will kill us, it feels like open AI is perfectly happy to just run down the road as fast as it possibly can, no matter what.
15:04100%. So this reasoning model, though, just keeps throwing me for that exact reason. Like, why, if you're open AI, launch this model? Even open AI did its own testing. It's a medium risk, right? Like, the company has, they've talked a big game for years about wanting to do this safely and wanting to take their time. And it's a complicated moment in the AI space and all this stuff. Like, what's, cycle analyze Sam Altman and co here for me for a minute. Why do this right now? Well, Altman said this week at a conference, he said he likened the capabilities of O1 to GPT-2 in 2019. So it's not that capable of their models.
15:48I think releasing it, my jaded side is that it's capitalism. They need to raise this money. They need something to show for it. Like we're asking for more money than anyone has ever asked for. So we need something to show for it, even if it's a half-baked reasoning model. And their position is also we need user feedback. We need the data because we can't just do this in-house forever. We need to release it at some point so we can see how it performs and how it can be better. so that that's their position but yeah it's a to me it's a funding and a competition standpoint you know everyone else is going to be coming out google has already claimed that they have reasoning models it's just it's the next stage for what they're gearing up to do which is uh quote-unquote agents yeah walk me through a little bit because i think my next question was going to be why raise this much money right now and on the one hand you might as well right like that's sort of the story of the tech industry is like if somebody will give you money, you mostly take it.
16:51But it also does seem like I have a bunch of thoughts about this like specific moment we're at in the sort of AI productization universe. But as you're saying with the agent stuff, it feels like there is a thing that's next. And is this six and a half billion dollars to get to that thing? Like, Do you think it's as simple as that? Well, yes and no. But yes, they have five levels to AGI. So it's chatbots, number one, reasoners, which is the reasoning. Agents, number three. Four is innovators. So these agents can help with inventing cures to diseases. And then number five is organization, so they can do the work of whole companies.
17:34This is something they laid out very recently. I just want to say, by the way, whoever came up with that, genius. like the way you just described that sounds so reasonable and it makes sense and i can sort of see how one flows into the other and then it's like as soon as you start to pick it apart it all just becomes insane but like totally as as a copywriting exercise like crushed it great job open ai precisely yes and i explained this and this is something i've argued with readers about i explain this not because i think that we are there yet or on our way to agi but because this is what they genuinely believe they're building like they think that they're building god i think that's important to lay out.
18:09So agents is, you know, I think it means different things to different organizations. For example, Salesforce built Dreamforce as the biggest AI conference in the world. And they called it Agentforce because only a few weeks... I hate that. And I was there all week. There's this fabulous photo of me demoing this with Benioff for hours. It's basically a Slack bot that will create a customer service bot for you. And they're billing it as an AI agent. So customer service chatbots are one component of agents, right? So replacing repetitive tasks. For OpenAI, the repetitive task they want to replace is everything, every repetitive task you can think of.
18:58People doing things. People doing things. So they're hoping reasoning can help agents make more informed decisions and do your life for you. This is what everybody wants, right? Like this is now the path, it seems like. And I mean, this is like Rabbit was talking about this at CES this year, right? Like that is what everybody wants to do. And I think to some extent that as an end point makes sense, right? Like I think Like there is no evidence that anyone is building God at this moment in time. But the idea that we are, that they are on a real path to building things that can do things for you on your behalf.
19:36Yeah. Seems real, right? Like I think we've been on that path for longer than people want to say. Like you go back to like the Google Duo thing where it would call a restaurant and make you a reservation. Yeah. That's like an old school communication mechanism, but that's the same idea. Yeah. And it does feel like, you know, you talk to Google and you talk to Apple about the Apple intelligence stuff. And like, that's the idea of all of this is go execute tasks on your behalf. And I guess what I wonder is, is there any particular reason to believe OpenAI is going to get there before anybody else?
20:12Like, I've become, this is the other thing I want to talk to you about. I've become totally obsessed with this thing that seems to have happened where Anthropic, a few months ago, released Claude 3.5. And all of a sudden, everyone I know who uses AI tools is like, oh, this is the best one. It's more fun. It's more interesting. It's more creative. It's like, this is just the best one. And it just happened. And all of a sudden, OpenAI went from being like the Goliath in the room to everybody being like, their tech is cool. But there are a lot of companies that make cool tech, and this stuff is being commoditized.
20:43And AI seems cool, but do we need OpenAI to be the sort of harbinger of all of this? And then OpenAI is like, we're raising all of the money in the universe at the highest valuation in the history of the universe. Here we are. And I just, I can't figure out what it is that is still so sexy about OpenAI in this moment. I mean, no one has the users for their chatbot like OpenAI does. Fair. That's a fair point. I think that's what I think about when I think of Anthropics chatbot. And, for instance, Grok, XAI's chatbot. So, you know, that is just— I try not to think about Grok, generally speaking.
21:20Grok sometimes appears, and that's okay, but I don't—yeah. I just leave Grok over there. Unless you need the president's holding a bomb, like, we don't really need Grok yet. Right. So, in terms of mass adoption— It has its use cases, yeah. Exactly. Exactly. So, yeah, I think that ChadGPT is a household name, and I don't think that's going to change anytime soon. And that's, you know, good for them. I think that any company right now is looking for a way to justify why they're spending millions and billions of dollars on compute or like AI research. They need something to show for it. And I think agents is the next thing they're reaching for.
22:02Whether they're going to get there or not, whether these are going to be useful, really remains to be seen. But that is all what they're working on because they see a product that could justify why they're spending so much time and money on this. That's fair. Is it possible that SearchGPT is that thing for OpenAI? Like, that's not quite an agent, but it does feel like I totally buy the case that ChatGPT is not a killer app for anything. I've been saying this forever, and I feel very vindicated by it because people are starting to come around. And a chatbot is not the future of computing. It just isn't.
22:36It just isn't. And I think a thing I've started to hear from more and more people is that OpenAI is a company that's really good at technology and really bad at product. and so I buy the theory that like if you can do agents well and first, that becomes a thing that it's like, oh, now you've built something I can use that is more than just like a novelty computer to talk to. It's like actually useful tool on my behalf. Still kind of feel like everybody's working on that. But I also think there's a non-zero chance that that thing might be search GPT, which they announced and then sort of seemed like just stop talking about.
23:14Do you use perplexity more than Google? No. Got it. Yeah. So I don't think so. I don't think anyone's going to be using AI Power Search more than they're just using Google and what's built into their devices, which is why there's huge antitrust lawsuits and such. Well, yeah. That might change. Yeah. But, yeah. No, I don't see SearchGPT being their next killer thing. Okay. We're all forgetting about Sora, throwback Sora. Oh, yeah. I did forget about it. I straight up did forget about Sora. Okay. Sora being the video model. That is very good slash horrifying. I think they're all, and OpenAI included, are, I think OpenAI bills itself as more, you know, research focused.
23:54Like, these are just the toys to show off their models that are so powerful. But they need to make money. And hence why they are not going to be a nonprofit, according to reports. Like, the agents, I think, are going to be what they're hoping works best for them. and yeah, humanizing them early makes a lot of sense in terms of the product vision. Yeah, that's fair. Do you actually think OpenAI is happy being like infrastructure kind of enterprise-y company? Like there is a version of it that OpenAI decides, you know, we want to be, I don't know, we want to be AWS, not Amazon.com, right? And it turns out that's a really good business and it's not sexy, but you don't get hauled in front of Congress and you make all the money in the known universe.
24:40And OpenAI could go that way. It could be developer-focused. It could be the back end of everything. It's kind of doing that successfully already. But I just can't shake this idea that that's not actually what OpenAI wants. No, I don't think so. Just an infrastructure company. Again, I think it can't even be compared to Amazon or AWS. They think that they're building God. I do think Andy Jassy does not sit around thinking about building God. Exactly. I think that is fair. I'm you've had a very long and successful career. What is a company you can compare like that they thought they were building God?
25:14Every NFT company that existed. No, I think I think that's right. And I think the it's been the strangest thing for me about trying to cover this stuff is I have never encountered a group of people who earnestly believe something that big is right around the corner. like even people 25 years ago when i was not covering this stuff because i'm not that old kylie uh who were talking about like the internet was going to change everything they weren't talking about god like there was no question of the new a new human relationship with a an entity more powerful than us like that thing is new yes and i don't think it's real to be clear again i don't think there's any evidence that it's real yes we're going to be real anytime in our lifetimes But a lot of people really sincerely believe that is what they're working on.
26:05And I still have not wrapped my head around what to do about that. It's very strange. And it is true. If you believe you're God, if you believe you're building God, you're sure as hell not building AWS. us. Exactly. Exactly. I have talked to like the extreme optimists and the extreme pessimists regularly. And the only thing I know is that this is the most stressed I've been covering any technology ever, because it can be argued forever and ever. The people who are building it genuinely, yes, believe they're building God and that this is all possible and we're going to reach AGI. And I asked someone, I was at a Time magazine dinner on Monday, and I asked a really important technologist that I'm not sure if I'm allowed to name, I asked them, do you believe in your heart or do you believe in your wallet that you are building AGI right now?
Read the full transcript
26:53And he was like, I genuinely believe in my heart that that is what we're building. It's a level of delusion I can respect. And then, you know, the pessimists think that this is just going to be a complete nothing burger. It's just going to implode, which I get. So listening to both these points and it coming down to sort of a philosophical argument is, is this thing thinking? Is it reasoning? It comes down to that every day. And it's wild. It's just insane. Yeah. I just want to say, for the record, robots, if you're listening, Kylie said delusion. I didn't say delusion. I think you're terrific.
27:31Don't get it twisted. So we're about to enter this really interesting phase for the rest of the year, right? I think MetaConnect is coming. We're going to hear a bunch of AI stuff from Meta. Gemini is still chugging along. Apple Intelligence is sort of kind of starting to a little bit ship. Does it feel like we're heading into kind of a new moment of AI? It's the agent stuff. It's all of this. Like, it sort of feels like if ever this stuff is going to go legit and mainstream, it might start now? You know, I am impressed with OpenAI's ability to help me choose my medical insurance. That happened recently.
28:15I just turned 26, and so I needed to figure out medical insurance. And it helped me choose, based off what I was concerned about, the best medical insurance. Did it give you like a real company that exists that you... I gave it a PDF of my options. So I was like, I don't understand all this jargon. Can you please break it down for me. This is what I need out of the insurance. Which one should I pick? And it was helpful. So its ability to parse information can be really helpful. I am not fully convinced further than that because it's so much hype and so much marketing and so much money involved that I just don't see us reaching this super intelligent model that is capable of reasoning on my behalf and taking over my laptop and doing tasks for me.
29:01I don't see us reaching that anytime soon. But I just talked to like a safety researcher who who thinks that I'm completely wrong and that in three months that I'm going to regret thinking this. So I don't know. I'm just not fully convinced yet. Yeah, that's fair. And O1 didn't push you any further in that direction. I thought O1 was actually really cool when I demoed it because I like being able to see how it is breaking down a problem. But no, do I think it's really intelligence or solving any really genuinely earth shattering problems? No, not yet. Yeah. And there's a certain amount of like showing your work is only impressive if you're still ending up at the right answer.
29:44Exactly. And I think it very much remains to be seen how often it is actually going to arrive at the right answer. Exactly. I have, I think, a bit of a higher tolerance than most about spending the time and money on figuring out if this is going to be good for humanity or not, I think that's okay. If it's fucking up and being weird, I think it's fine until we figure it out. But it can't be forever. And I think I'm more allergic to the hype and this is going to change the world and our SaaS product is going to make you 10x more productive. That I'm allergic to. Yeah, that's totally fair. just remember that crypto did change the world just the way everybody said it was going to and so did NFTs.
30:28So many good memes. And we're doing this on a Web3 platform right now. So everybody's right when they talk about everything that's going to change the world. Great point. That's what I would leave you with. Yeah, you're welcome. All right. Well, thank you for coming to my occasional existential crisis about AI. Kylie, we're going to do this a lot more this year. Now you're not going, there's no more Burning Man until next September. So now you have to come on the Vergecast sometimes. I'm going to start sounding insane. So good luck the rest of the year. I'm really excited. All right, we got to take a break and then we're going to come back and we're going to talk about some policy stuff.
30:56We'll be right back.
31:21and get qualified candidates that you can manage all in one place. LinkedIn's new features can help you write job descriptions and then quickly get your job in front of the right people with deep candidate insights. Either post your job for free or pay to promote. Promoted jobs get three times more qualified applicants. At the end of the day, the most important thing to your small business is the quality of candidates. And with LinkedIn, you can feel confident that you're getting the best. Based on LinkedIn data, 72 % of SMBs using LinkedIn say that it helps them find high-quality candidates. Find out why more than 2.5 million small businesses use LinkedIn for hiring today.
31:59Find your next great hire on LinkedIn. Post your job for free at linkedin.com slash track. That's linkedin.com slash track to post your job for free. Terms and conditions apply. Oh, hey, welcome to gift wrapping. Whoa, Zoe Saldana. Hey, can you wrap these, please? Wow, iPhone 17s. You splurged. At T-Mobile, you can get four iPhone 17s on them. It's the perfect gift for everyone. I'm the worst. I only got my mom a robe. Well, it's better than socks. So, I have to trade in my old phone, right? No, at T-Mobile, there's no trade-ins needed when you switch. Keep your old phone or give it as a gift.
32:36Incredible. In fact, wrap up my old phone, too, for my Aunt Rosa. Forget that. Aunt Liz will be jealous. Sounds like my family drama. Oh, I got it. I'll give it to my abuela. I'll take reindeer paper with, hey, where are you going? To T-Mobile! Thanks, Zoe! The holidays are better at T-Mobile. Get four iPhone 17s on us. No trade-in needed when you switch. Plus four lines for just$25 a line. With 24-month legal credits and four eligible portings on essentials, for well-qualified customers, auto pay plus taxes, fees, and$35 device connection chart. Credits end-in balance due if you pay off earlier.
33:07Cancel. Contact us. Finance agreement. 256 gigabytes. $830 required. Visit T-Mobile.com. Support for this show comes from Zapier. When people talk about AI, they often talk about the future. How will AI affect the economy? What will AI's potential be in the future? Will AI be able to make popular art? But the more important question is, what can AI actually do for your business today? If you're looking for concrete implementation of AI for your business, you can try Zapier. Zapier says they can help you and your team actually reap the benefits of AI, not just talk about it. With Zapier's AI orchestration platform, you can bring the power of AI to your work.
33:44Connect top AI models like ChatGPT and Cloud to the apps your team already uses, so you can add AI exactly where you need it. Whether that's automatically enriching leads, coaching your sales rep, resolving IT tickets, or more, you can do it all with Zapier. Zapier is for everyone, tech expert or not. No IT bottlenecks, no complexity. No AI hype, just results. Teams have already automated over 300 million AI tasks using Zapier. Join the 3.4 million companies already automating with Zapier and transform how you work with Zapier in AI. Get started for free by visiting zapier.com slash verge. That's Z-A-P-I-E-R dot com slash verge.
34:35All right, we're back. let's talk politics. But not like, not really. Like, don't worry. We're not going to yell about politics. But there are some interesting things happening right now in the kind of legal, regulatory world. And also, there is a politics thing happening. But I promise we're not going to spend too much time on that. I asked Gabby Del Valle and Addie Robertson on our team to come on and just kind of do a run through with me. We've got to talk about TikTok. We've got to talk about Google. We've got to talk about crypto. We're just going to blast through all of it, See if we can catch up.
35:08And then we don't have to talk about politics for a minute, which personally makes me very excited here on The Verge Cast. But sometimes, sometimes we got to do it. So let's do it. Addy, welcome back. Hi. Gabby, hello. Welcome to the show. Hi, thank you. First time, this is very exciting. I'm thrilled. I can tell. That was the enthusiasm I was looking for right there. Well, I smiled. They can't hear that, but I did smile. That's just what my voice sounds like. There's always a little bit of sarcasm in it. It's all right. I'll take it. Okay. So I want to do three things while the three of us are here together.
35:43I want to talk about the TikTok hearing. I want to talk about Trump's crypto thing. I don't even know what to call it. We're going to talk about that. And then I want to talk briefly about where we are in the Google ad tech trial. And then we'll get out of here. Does that sound good? Great. Okay. Sounds good. Let's start with TikTok because if I'm being completely honest with you, I had like completely lost track of all this TikTok stuff. I feel like kind of like the last time they tried to ban TikTok, I just sort of forgot about it and then it went away and I felt very justified about that. But there was a big hearing and it appears we are we are still in the process of figuring out what's going on.
36:19So, Gabby, can you just give me kind of the rundown of what happened at this hearing? Yeah. So first, I want to talk a little bit about what happened before the hearing, which is that the government introduced a bunch of classified material in its filings. And they were like, the judge can look at this. We can look at it. TikTok can't look at it because it's a national security concern. so we can talk about it to a point but there is just some stuff that we don't know because it's like pages and pages of redacted material the actual hearing did not focus too much on the redacted material it was more about um the on the one hand tiktok saying that the government did not really consider all of the possible options and on the other hand the government being like well Well, actually, we did negotiate with you guys for three years, I believe.
37:11Adi, was it three years or was it two? So TikTok has been trying to talk about mitigating this since about, I think, 2020 is when Trump made his first we should ban TikTok proposal. And then they've been seriously in discussions since around 2022. That was when Project Texas happened. It started Project Texas in 2022. that was the period at which it was saying we can mitigate these problems by teaming up with Oracle, which, as you all remember, was also the company that was going to buy it during the period where it seemed like Trump was going to try to really force a sale. Yeah. So this I'm starting to feel vindicated again by my stance of just not really paying attention to this and assuming that it will be chaotic and then go away because all of that came to both kind of nothing and where we are now, right?
38:00Like, Gabby, that's kind of the run-up to how we got to where we are. We've been at this a long time now. Yeah, we've been at this a long time. And basically what the government is saying is like, we gave you a million chances. You could not do what we wanted. And TikTok is saying, you didn't give us enough chances, actually. We gave you this proposal. We gave you maps of our offices. We gave you all of these things, please. but the government's argument is also not that tiktok has done anything at the behest of the chinese government just like maybe one day it could so well and that's that's what's been so complicated about this whole thing right like adi you and i have talked about this before it's that it is this uh impossible to prove and impossible to disprove thing that it's like it it seems possible that something very bad could happen.
38:51And that has gone back and forth a million different times. And it does feel like we're running at the question of like how legally defensible is the idea that something bad could possibly happen in this case. A lot of which I think just depends on how threatening do we think China is? Because as long as a Chinese company owns TikTok, like no matter how much they silo it, there's just, I think, not technically anything you can do to prove that the company that owns your subsidiary could not at some point access your data. Right. Yeah. And so the sense I got just reading the coverage we did of the hearing was that it kind of landed in that same place, Gabby, that like it's just everybody making the same arguments that all kind of talk past each other.
39:39And it doesn't sound like we're in a place where there are all that many arguments that actually address each other? Because one side is like, it could be bad. And the other side is like, well, it's not bad. And the other side's like, well, it could be. And I don't know how you overcome that. Yeah, exactly. And also, I would say that the judge's role in this is really interesting. At one point, one of the judges said that TikTok was owned by China. And then TikTok's lawyer had to be like, no, no, no, we're owned by a company that has a holding company. I think he said in the Cayman Islands, he's like, we are not owned by China.
40:11And, you know, that's true. The Chinese government does not own TikTok. But the DOJ's entire argument is, you know, they could influence it. They could. Have they? We don't know. Probably not. Maybe. But they could. And in my opinion, I don't know what you think, Addy, but it seemed like the judges were kind of buying that argument. Like they were really focused on the national security risk of the whole thing. And the hearing, I think, wasn't even so much directed at exactly how much of a threat is TikTok. It was directed at if Congress thinks there's a threat, can you legally force a company to divest?
40:49It's not even like the question of whether it's a threat, I think, comes after all of that. Right. I guess that's true. There is kind of a meta aspect of this where we're not actually arguing about in this particular case. I think I have this right that whether or not the allegations about TikTok are true. It's the question of if Congress believes the allegations against TikTok, what is it able to do about them? Right, which feels sort of one step removed from the actual impossible question, but is ultimately at this moment maybe the only part of the question that actually matters in terms of what happens to TikTok.
41:27Right, and I think there are caveats like, are we at active war with China? Like, I think that those are just really gigantic mitigating factors, Like exactly how much of a threat do we think China is just as a legal status? But yeah, a bunch of this is really just can you prevent something that is a foreign government that we think that Congress thinks presents a threat from operating inside the United States as a commercial entity? So, Gabby, you were saying it felt like the judges were buying the argument. What was kind of the vibe at the end of the hearing? What were you thinking about how it went down?
42:01It was very unclear to me what would happen, but based on the questioning, I felt like the judges were both like torn between whether national security concerns kind of trump this First Amendment question, but also like they did seem to genuinely be like, well, if China is a threat and if Congress believes that China is a threat, then we have to take that seriously. whereas TikTok's entire argument was not only are we not owned by China we are not a threat and at one point they were like other companies and platforms based in China are not being singled out in this way so why are we being singled out in this way um which I do think is you know an interesting question well and it seems like with some hindsight now the answer to that question is still A, very mysterious, but B, has something to do with Israel and Hamas and what happened last October.
42:55And I feel like I got on this podcast and got very loudly upset about like, if you think there is something going on here as the government, you have a responsibility to tell us what's going on. And at least as far as I've seen, we haven't gotten any of that. But do we have any kind of new information or understanding about what is the root cause of all of this? Why everybody is so mad at TikTok and not mad at these other platforms like you're talking about? I think that Israel and Hamas is a big part of it. I don't think, as far as I'm aware, Congress has not proved that there has been any kind of influence campaign.
43:29But there have been several members of Congress, both in the House and the Senate, who have said that, you know, the campus protests, for example, show that our youth are being influenced by malign foreign actors, that they could not have possibly come to these opinions on their own, and that there is somebody putting this information in front of them. In one of the declarations that was filed by the government, there was mention of a feature on TikTok's back end called heating, which is when you can kind of just boost certain content if it's trending or if you want it to be trending. And again, it was said like, we don't have any proof that this has been used maliciously, but could have been right yeah i the the biggest thing i remember is there was this moment i think it was last fall maybe early this year i don't know i've lost track of time uh where there was a there was an intelligence briefing and a bunch of congress people came out and voted unanimously to ban tiktok and and the overwhelming question has been like what did they learn in that briefing And it still feels like whatever it was, we don't know.
44:35And at this point, it's almost shocking to me that we don't know. Yeah. And if you look at the filings, like big chunks of it are redacted. And I guess, you know, if there's valid national security concerns, I do get that. But it's also like, what's in there? What did they tell you? Right. And especially like it's if there are sort of ongoing actual national security concerns, sure. But if it's like heating is a feature that exists, like maybe maybe unredact that. Like, it's fine. Let us let us know what's going on here. Well, that part was was unredacted. I mean, let me pull up the filing. There was one part that I remember being finding very interesting.
45:10Please. Right. OK, so there's a section in one of the filings, eight pages long, titled ByteDance and TikTok's History of Censorship and Content Manipulation at PRC Discretion. Almost entirely redacted. What is that history? I don't know, but it's in there. Maybe. And the fact that it's redacted makes it sound like there must be there must be something. So in that same declaration, it's from Casey Blackburn, who's an assistant director of national intelligence. He writes that there is, quote, no information, end quote, that the Chinese government has used TikTok for, quote, malign foreign influence targeting U.S.
45:47persons or the collection of sensitive data of U.S. persons. Just there's a risk of it happening in the future. Right. And yet we're being very secretive about all the stuff that's being discussed. I just I cannot square those two things in my head. Adi, can you square those two things? This fact that there is a lot of stuff being redacted. We're having this big semi-private sort of obscure debate about what's going on. And then even the people in Congress are saying there's no evidence that this stuff has been going on. How do you make sense of those two things happening simultaneously? I don't know.
46:17I mean, part of the question is how sensitive are the things that are actually redacted? I mean, we understand that there's a very, very long history of sort of over-redaction and over-classification in the U.S. government. And so it's plausible that these things like they are sensitive, but they are also not necessarily incredibly divergent from the things we've heard publicly. But on the other hand, I don't know. I mean, I think the question really comes down to something that is still just not related to that, which is what is the First Amendment right that you have to operate something as a company that has a foreign owner?
46:53Right. Yeah. So we should pivot away from this. But what what is the next piece of this? Because we're still we're like barreling towards January, which is theoretically the deadline for TikTok to either be sold or banned, depending on who you are and how you read what's going on here. Is there a is there a next step between here and there that we know is coming? The next step is whether the D.C. Circuit Court decides that it should block this law, basically. And then at that point, TikTok, I think, correct, has until January to at least start the process of getting divested. Okay. This is like definitely going to the Supreme Court, right?
47:32That feels like the inevitable end of this process one way or another. Yes. Yeah. Okay. All right, cool. Just checking. So we'll be back. We're going to have many more bites at this particular story. Now I want to talk about the story that I just don't understand and basically didn't pay any attention to. But Gabby, you in particular had to. And I just feel like I feel like we we owe it to you to let you talk about your feelings about this. Tell me about the Trump space from from last week. Oh, my God. So the way I've described it to Addy in numerous times is like a jigsaw torture scenario designed for me specifically.
48:10it was you know so let's rewind a little bit Trump was the headliner at the Bitcoin conference this year he's been really trying to you know show the crypto community that he's behind them he wants their votes etc etc and his sons and him have been teasing this crypto platform World Liberty Financial. And he was supposed to announce it in a Twitter space at 8 p.m. Eastern. So, you know, I logged on to the Twitter space at like maybe 7.55. Addy was there, too. I didn't suffer alone. And I from 8 p.m. Eastern until what, like 10.30? It was about 10.30. It was about 10.30. Nobody announced actually what it was.
48:56But they started talking about like some of the details, but they never said just what it was. And I was doing my pre-write like for it and I was taking notes. But my it was just like Trump announces TK and then he got off the stream. So I had to change the whole thing. He didn't announce anything. He was talking about how good his granddaughter Arabella is at speaking Chinese. He was like she impressed she so much. He loved her. He loved how she spoke. And I was like, what is this crypto platform? Please. The best way that I had to describe it as like imagine an apple event or like a game announced at e3 and steve job comes up and he's like i'm announcing something and then he doesn't actually announce the iphone and they just start all talking about the iphone like like features of the iphone he's just like here's how you cut and paste and you're like on what exactly that's very good so okay so let's do some work for this Twitter space.
49:54What do we think this thing is? To be honest, I'm less curious about the product itself and more about sort of what it means in this political moment in the United States. But we should attempt to define the thing a little bit. So what is this thing that they kind of sort of almost didn't a little bit launch? So still actually unclear, but three people with knowledge of the project told the New York Times that it has been pitched as a borrowing and lending platform. So to be clear, they did not tell the New York Times that it is a borrowing and lending platform. They said that it has been pitched as a borrowing and lending platform.
50:31I like that distinction because maybe it's not even that. I don't know. Maybe it is. But the thing that all of them kept talking about, less so Trump because he was just talking about, you know, his usual stuff. But Donald Trump Jr., some of his business partners, they kept talking about debanked people and underserved communities and how this is really going to help out a lot of vulnerable people who have been shut out of traditional financial markets. And as Addie pointed out when I first wrote about this, she was like, remember Trump university remember that so trump university was if i remember correctly um it was before trump ran for president it was you know his apprentice times it was a course program where you could learn to be a real estate investor um but then there was a trial because a bunch of people sued him over how the point of trump university was not to teach people to invest in real estate it It was to sell them more courses about investing in real estate and other business activities.
51:40So it was like basically a scam. It was basically a scam targeted at people who would like to get rich, like Donald Trump. Okay. And then, Addy, what did you make of all this? So the actual detail that we have based on the stream is that, A, it is supposed to, not an exact quote, but a paraphrase, drive mass adoption of stablecoins and be easier to use for normal people. Stablecoins, of course, being like coins that are pegged to an actual currency. And it is pitched specifically, so there is unbanked people. There are unbanked people. Those are people who do not have a bank account. They tend to be people who do not have very much money.
52:19It is a genuine issue. And those are folks that the crypto community has been talking about forever. That is like if you want to make like a big, beautiful, good for the world case for crypto, those are the folks that everybody starts with. And it's mostly that's been tested overseas and the results have been somewhat mixed. This is more specifically they pitched to debanked people, which is basically people that have been quote unquote canceled and they lost their bank account because of it, which is like, let's just say a much more niche group. Yeah, I feel that laughing when you say that. I just assumed that debanked and unbanked were like synonyms for the same thing.
52:57It is deeply upsetting that that's not the case. Oh, no. There was a really great quote from Donald Trump Jr. who was like, you know, there was a time where the Trumps, we could have picked up the phone. We could have gotten any CEO of any bank and gotten a loan from anyone in the world on any project. And then we got into the political arena. And then he said, we went from being people who would have been the elite in that world to just being like totally canceled. And it's like, well, you also your dad got sued for fraud. So maybe it's not. I don't know. It seems like it fits really well into the sort of alt-tech ecosystem.
53:35Stuff like the Freedom Phone and various alternative social networks. Things that run from they actually exist as product to being basically scams or rebadged Android phones. Yeah, I mean, and I think the reason I'm interested in this is less about the actual product itself And more about, A, that ecosystem, which is real and growing, and there are a lot of people making a lot of money from that ecosystem, some of the real products and some of the scams. But also the fact that so much of what the tech industry has been talking about with respect to this election has been either explicitly or implicitly about crypto.
54:18Like in a very real way, there is like a dividing line in this election and it has to do with how you feel about crypto, which I think in a lot of ways has to do with how much money you have invested in crypto. And my read of this whole thing was it was just this entire project is just another way for Trump and the Trump campaign to say we love crypto. Am I misreading that or understating what they're actually trying to do here? I think that's part of it. But I think another big part of it is that they're really appealing to the victim complex that a lot of people who are really into crypto have.
54:51When I was at the Bitcoin conference, there were all these people talking about how terrified they are of like central bank digital currencies and how, you know, eventually the deep state will just be able to like completely shut out whoever they want out of the financial system. And that's why crypto is so powerful. It was it was an extremely political or politicized environment at the Bitcoin conference this year. It wasn't just Trump. I mean, Cynthia Lummis spoke. Edward Snowden eventually gave a speech that was like, yeah, all these politicians are making promises to you, but maybe maybe you don't trust him.
55:22And it was really funny because he got a standing ovation before he spoke. And then he was like, I can't see you guys. I'm like like web calling in right now. But thanks. And then he didn't get a standing ovation afterwards. afterwards. And I remember talking to a friend and being like, do you think that's because of what he said or because he they know that he can't see them? And she was like, definitely both, like, but probably more what he said, because he really like threw cold water on the whole thing. He was like, maybe it's not great that there's a bunch of politicians here. Maybe that's bad.
55:54Yeah. I mean, it is like the whatever last five years of the crypto movement. I think if you ask people not that long ago how it would feel to have a bunch of politicians be the biggest names in crypto, they would feel very differently than they seem to today. I think that has changed in a pretty big way, even just over the course of this election cycle. All right. No more crypto. Let's just talk ad tech in other deeply exciting things. I just want to check in on this trial. Lauren Feiner on our team has been in the courtroom, I think is probably in the courtroom right now as we're doing this.
56:26She's not in the courtroom today because she is traveling. Okay, that's good. Poor Lauren. I was there for one day, and I had to sit in the courtroom. There are no devices, no electronics allowed. I couldn't even wear an Apple Watch, like nothing. And I had to sit there with a notebook and a pen like it was the 1700s and listen to people talk about header bidding and yield management. And it was really something special. But Addy, you've been editing and assigning a lot of this coverage. Lauren's been in the courtroom a ton. I've been in the courtroom a little. It feels like we're getting to kind of the end of the government's part of this trial.
57:04What's your sense on where we're going and what we've learned so far? Correct. The DOJ wrapped up yesterday. We're expecting Google to be making its case through mid-next week. It seems like we could have a wrap of this part of the trial. It's not a sure thing yet, but the end of next week. Okay. And so far, it's hard to say what the judge, how the judge is feeling. That's always very hard to say. And I think we haven't gotten a really clear picture in this case. But so far, the DOJ's case is basically Google has incredible power, which I think most people don't dispute, and has gotten it by making all these very hard-noticed decisions that consolidate its power in a way that relies on the fact that it has access to all these different parts of the spectrum.
57:47It has what was, at the time of most of these events, DoubleClick for Publishers, which was a Google's publisher ad sales server. And then it has access to the sort of other side of the equation for advertisers. And then it has AdEx, which sits in the middle of this. And a lot of what ended up coming up is the idea that it used AdEx access and access to all this very high quality data about the advertisers to strong arm publishers into taking deals that they didn't really like and into neutering projects that and features that might help people diversify away from Google's tools. It's been really interesting to follow this trial after covering the search trial a bunch last year, because so much of the overarching case here kind of rhymes, right, in the sense that Google is like, yes, we're very big and we're very successful.
58:45Like, correct. We won. We did it. Good job, us. But it's because we're good at it and because our product is the best and because people like to use it. And the government is making almost the exact argument they made back then, which is like, no, you got big because you were good. Everybody like, yes, granted, congratulations, you did it. And then you spent years or decades just ruthlessly preventing anyone else from ever being able to do the thing that you did. and like the specifics and the details and the tech are so different. But the case is just the same. Google is like, we win because we're good.
59:23And the government is like, well, you're not actually that good anymore, but you keep winning. So what's that about? Right. Which is how they demonstrate the idea of harm, which is that, yeah, Google made all of these products that people thought were really good. And then they created a situation in which the DOJ alleges the market stagnated. And that it's a little bit harder because obviously anybody can use search. You can look at search. You can decide for yourself whether you think search is still good or not. In this case, it's a little more indirect, but a lot of the idea is people ended up paying more for ads.
59:54It was harder to run a business that was ad supported. It ended up making, say, ads maybe more intrusive because you have to just spam people with a bunch more. And so there's the DOJ isn't making that case nearly as much. But for the average person, that is sort of more of the takeaway. Yeah, it is. deeply, deeply wonky in every possible way. And even talking to folks like in the ad tech business about all of this, their eyes kind of start to glaze over when you talk about like the actual underpinning technology of all of this, where it's like, oh, let's talk about publishing servers and the lock-in effect that they give for the rest of the, and everybody just like falls asleep at their desk talking to me on Google Meet.
1:00:35Which I was very impressed with the judge in particular the one day I was in court and in her ability to just stay locked in. And she was asking a lot of like vocabulary questions because everything in this entire case is just insane acronyms that no person should know. But everyone in the room just assumes, you know, because they all do it for a living. And so at one point she was just like, sorry, what's what's torso and tail in ad tech? She just like paused the whole trial. She's like, what does that mean? Why aren't you explaining this to me? It was great. It turns out it means medium and small publishers.
1:01:07The head is the big ones. the torso is the mid-sized publishers and the tail is all the like cooking blogs and single person operations that are less the focus of this trial. I thought torso was an acronym. I'm so sorry to interrupt this learned conversation, but I thought you were going to be like, I don't know. I can't even think of the letters that torso would be, but I did think it was an acronym. The optimal return sick openness. Yeah, that's what it is. You're right. That's one of Google's products. Not a lot of people know about it. But no, and I think you're kind of right, Addy, that it seems like it has been really hard to figure out how any of this is going for folks.
1:01:47Like reading our coverage, reading other coverage, the government seems to be making a case. Google seems to have strong thoughts about every part of that case. And then the strangest part is like the search trial, it's just down to this one judge who is just sitting there kind of quietly listening and asking vocabulary questions. What's going to happen? And I think it would be so interesting if this was a jury trial. And Google obviously wrote a big check to make this not a jury trial. But it does feel like it would be different if it was in front of 12 random people off the street instead of one judge.
1:02:21Yeah. So far, I think the clearest thing and the thing that seems most likely to go badly for Google is the same thing that was an issue in the search trial by Google overclassifying and not storing chats and turning off chat history and things like that. It seems not necessarily clear that's going to play in hugely into this trial in the way that it didn't last time, but certainly seems like something where the judge is not incredibly sympathetic. Yeah, Google keeps doing the thing where they're like, oh, let's talk about this. And then somebody goes, oh, chat history is on. And then it's like end of chat.
1:02:53Like, oh, well, that's not incriminating at all. Like, that's that's cool. Which is like it's not necessarily incriminating. Like, obviously, but in the last trial, there was a bunch of, OK, you're just saying things that I understand they're not necessarily incriminating, but they also kind of make you sound bad. Right. Exactly. They just make you sound really Machiavellian. Yeah. There's something about saying nobody listen to this that just kind of makes it seem like you're about to say something you don't want people to hear. Like, it's hard to it's hard to argue with that. And even the stuff when they weren't doing that, it was like, OK, well, maybe we should be clear that we use we shouldn't use this terminology because that'd make us sound like a monopoly.
1:03:27Right. My I Am Not a Monopoly shirt is raising a lot of questions already answered by my shirt. Yeah. I'm going to do you think they'd let me wear that shirt into the courtroom? I'm going to go. I'm going to make that shirt and see if they'll allow me into the courthouse in Alexandria and see. And I will report back. It's going to be great. All right. We got to take a break. I should let you guys go. But I think it's it's about to be an election and we're going to have to have you both back. And we got some more stuff to talk about between now and then. But thank you both. Oh, God, that is happening.
1:03:55Yeah. All right. we got to take one more break and then we're going to do a question from the Vergecast hotline. We'll be right back.
1:04:04Support for this show comes from LinkedIn. When you're a small business owner, your business is on your mind 24 seven. So when you're hiring, you need a partner that works just as hard as you do. That hiring partner is LinkedIn jobs. When you clock out, LinkedIn clocks in. LinkedIn makes it Easy to post your job for free, share it with your network, and get qualified candidates that you can manage all in one place. LinkedIn's new features can help you write job descriptions and then quickly get your job in front of the right people with deep candidate insights. Either post your job for free or pay to promote.
1:04:39Promoted jobs get three times more qualified applicants. At the end of the day, the most important thing to your small business is the quality of candidates. And with LinkedIn, you can feel confident that you're getting the best. Based on LinkedIn data, 72 % of SMBs using LinkedIn say that it helps them find high-quality candidates. Find out why more than 2.5 million small businesses use LinkedIn for hiring today. Find your next great hire on LinkedIn. Post your job for free at linkedin.com slash track. That's linkedin.com slash track to post your job for free. Terms and conditions apply. Support for this show comes from Darktrace.
1:05:18Defenders and cybersecurity are always there when we need them, keeping the systems that run our daily lives safe from bad actors. At Darktrace, they think defenders everywhere deserve a bigger spotlight for keeping the lights on. They deserve blockbuster movies about breaches that never happened. And they should get a parade every time they block a novel threat. But most of all, they deserve AI cybersecurity that can stop novel threats before they become breaches. across email, clouds, networks, and more. That's why Darktrace is going beyond for cyber defenders, such as IT decision makers, CISOs, and cybersecurity professionals to give them the power to see across their entire attack surface and stop the threats other platforms miss.
1:06:03Darktrace is the cybersecurity defenders deserve and the one they need to defend beyond. The world needs defenders. Defenders need Darktrace. Visit darktrace.com slash defenders for more information.
1:06:41a different kind of vibe coding tool. With just a few prompts, you can turn any idea into a code-backed prototype or app that anyone can explore, iterate on, and share. And it's all connected to Figma. So you can start from scratch or from your design file. Figma Make is made for designers, product managers, and builders who want to move fast and explore more ideas. Try it now and see how quickly you can make your ideas real with Figma Make. Just go to figma.com slash vergecast. That's F-I-G-M-A dot com slash Vergecast.
1:07:41you get your iPhones and try the new stuff. I want to hear what you're thinking. Hit us up. Tell me everything. This week, we have a question that is near and dear to my organizational heart. Let's hear it. Hey, my name is Erica. I'm reaching out with kind of a random question. You just went through your mini series on productivity, but the reason I don't find myself to be people productive is my digital debt. I have so many documents and so many photos. And anytime I try to clean them up, it just feels so tedious. And I'm sitting on a plane and I'm deleting screenshots and it's just really the best piece of my time.
1:08:19So if you guys have any ideas or anything you could think of, I would love it. Thanks, guys. Bye. All right. So I bring this question to you for a couple of reasons. One, because if there is a perfect answer to this question, and I have not found it. The idea of digital life being about just collections of stuff is super duper real and bad. So if there is a perfect solution, if you've devised a way to, A, not let your computer get super cluttered, or B, clean it up really quickly, let me know. I want to know. I have this newsletter called Installer. I will talk about it there. I'll talk about it here.
1:08:56I will sing your name from the rooftops. Tell me everything. But I do have a bit of a system for this and just figured I would share the way that I do it in the hopes that it might help. It's kind of three separate things. The first step is about your camera roll on your phone, which I think for a lot of people is where a lot of the file clutter exists. You take screenshots, you have pictures of your receipts, you have duplicate pictures of your kids and everything. I do two things here. The first thing is on both the iPhone and Android, there is a way to quickly delete duplicates on your phone, which I highly recommend and think everybody should do.
1:09:34It's just in settings in the in the photos app of both of those things. You can just delete duplicates. The other one is you can sort out screenshots. You can do it in Google Photos. You just search for screenshots. I'll find the screenshots. Apple pulls it all into a separate album. I periodically just go in there and delete them all. I have never once been burned by this where I'm like, Oh, I wish I had that screenshot again. Go with God, I suppose. There might be problems if you do it in such a sort of brute forcey way. But that is a thing I do every few months. And I'm always amazed at how many screenshots I have in there and how few of them I actually care about.
1:10:10Most of them are just like my lock screen. Anyway, the other one is to find an app that essentially lets you like Tinder swipe your way through your photos. There are a bunch of these. There's one called swipe and delete. There's one called swipe wipe. There's one I think called slide box. There's a bunch of these out there. Most of them will make you pay an ongoing subscription. But what I usually do is either just do the free trial or just pay like once and use it once. And essentially what they do is they show you a photo and you swipe right to keep it or left to get rid of it. And it is shocking how quick a way that is to go through some of your photos.
1:10:43Most of them, it's it's an obvious snap decision, right? Keep or lose. and having just a mechanism that lets you choose that is great. So that's the first thing, camera roll. Thing number two is your computer. And this is where my system is not great, but it really works for me. So I figured I would share. The first thing I do is use an app on the Mac called Disk Inventory X and on Windows called Winderstat, W-I-N-D-I-R-S-T-A-T. And what both of those do is just visualize the storage on your computer. They'll show you where all the big files are, where all the folders are with all the big files.
1:11:20And I'll just go in and find the biggest stuff that I don't need. A lot of it for me is like video recordings or huge application files that I download, caches from browsers, and just delete all of that. And most of the reason I think to do a decluttering like this is because you're starting to run out of storage. And that is a super duper efficient way to do it. Again, I think for a lot of people, photos are a thing that takes up a lot of space. So if you have photos elsewhere, you can delete them from your computer and free up a lot of space. So yeah, one of those visualization apps is an incredibly useful way to just figure out where your storage is going and start to get some of it back.
1:11:57The second thing I do is once every six months or so, I will just take all of the contents of my downloads folder and my desktop folder and for me, my documents folder and just put it all on an external hard drive. Again, I've deleted the biggest files by now. So most of what's left is like little things I downloaded and documents I made and whatever. But the way I work, at least anything that I'm going to need permanently ends up somewhere else. Usually it's in Google Drive, but sometimes it'll get filed to like a specific spot on my computer. Anything that is in sort of those like stock folders is probably something I don't need again.
1:12:36But I don't want to do the work of going through and actually figuring it all out. but I also don't want to delete all of it in case there is something that I need. So I have a little tiny two terabyte hard drive that just sits here on my desk. And every few months I plug it in, I empty the folders onto that drive. And then I don't think about it again. It cleans up my computer. It makes everything very simple. And if all of a sudden I'm like, oh, where is that thing? It's on the hard drive. Overwhelmingly, it is on the hard drive. That has been incredibly useful for me in both keeping my computer and me sane, but also not having to dedicate like four full days a year to keeping it that way.
1:13:15So that's my system. It has worked very well for me. I'm sure it's not the best system. I also am like relatively organized otherwise. So again, like things that I know where I'm going to need end up mostly getting filed somewhere. So if you're just kind of a like let chaos reign and then once a month tame it person, you should probably keep doing that system rather than mine. But I just wanted to share in case that helps. Start with the camera roll. get rid of the big files, dump everything else onto a hard drive and know that it's there if you need it. I hope that helps. That's all I got. And if you have a better idea, I would love to hear it.
1:13:49I think this is a problem everybody has. I'm looking at like the mass of icons on my desktop right now and I'm like, I need to do this again. And I think a lot of us do. So I'd love to hear from you. Anyway, for now, that is it for the Vergecast. Thank you to everybody who came on the show and thank you as always for listening. There's lots more on everything we talked about from TikTok to Trump to Google to OpenAI, all of it on TheVerge.com. All of this is also like ongoing, so we're covering it as we go. It is a busy season in the tech world. I'll put some links to all that stuff in the show notes, but keep your eyes on the website.
1:14:21It's a fun website. As always, if you have thoughts, questions, feelings, or other things that are gonna make me have an existential crisis about AI, you can always email us at vergecast at theverge.com or call the hotline 866-VERGE-11. I love hearing from you. It's the best. Thank you to everybody who reaches out. It is the absolute most fun Slack room we have that pipes all the voicemails in, and I love hearing from you. This show is produced by Liam James, Will Poore, and Eric Gomez. The Verge Cast is a Verge production and part of the Vox Media Podcast Network. Neelai, Alex, and I will be back on Friday to talk about a bunch of new gadgets, all the stuff going on at MetaConnect, and everything else.
1:14:55We'll see you then. Rock and roll.
1:15:07Support for this show comes from SC Johnson. We've all been there. Choosing not to wear your new white shoes because there's a 10 % chance of rain. Bending awkwardly over the tiny coffee table to enjoy a sip of your latte. Not ordering the red sauce. Those feelings of dread are what we call stainxiety. But now you can break free from your stainxiety with Shouts Triple Acting Spray that has stain-fighting ingredients to remove a huge variety of stains so you can live in the moment and clean up later. Just breathe and shout with Shout Triple Acting Spray. Learn more at ShoutItOut.com.
1:15:48Mercury knows that to an entrepreneur, every financial move means more. An international wire means working with the best contractors on any continent. A credit card on day one means creating an ad campaign on day two. And a business loan means loading up on inventory for Black Friday. That's why Mercury offers banking that does more, all in one place, so that doing just about anything with your money feels effortless. Visit Mercury.com to learn more. Mercury is a financial technology company, not a bank. Banking services provided through Choice Financial Group, Column N.A., and Evolve Bank & Trust members FDIC.
From the publisher
Kylie Robison joins the show to talk about OpenAI’s new model, o1, and what this new “reasoning” model says about the state of the art in AI — and what AI companies are willing to put up with in the name of building God. Then, Gaby Del Valle and Adi Robertson talk through the latest on the TikTok ban, the Trump crypto chaos, and the ongoing adtech antitrust trial against Google. (All with as little politics-talk as possible.)
Further reading:
OpenAI releases new o1 reasoning model
OpenAI’s new model is better at reasoning and, occasionally, deceiving
TikTok ban: all the news on attempts to ban the video platform
TikTok oral arguments will weigh security risks against free speech
TikTok faces a skeptical panel of judges in its existential fight against the US government
Donald Trump is hawking tokens for a crypto project he still hasn’t explained
US v. Google redux: all the news from the ad tech trial
How Google got away with charging publishers more than anyone else
Email us at vergecast@theverge.com or call us at 866-VERGE11, we love hearing from you.
Learn more about your ad choices. Visit podcastchoices.com/adchoices
