Deep Dive
Shownotes Transcript
Ryan Reynolds here from Mint Mobile. I don't know if you knew this, but anyone can get the same premium wireless for $15 a month plan that I've been enjoying. It's not just for celebrities. So do like I did and have one of your assistant's assistants switch you to Mint Mobile today. I'm
I'm told it's super easy to do at mintmobile.com slash switch. Upfront payment of $45 for three-month plan, equivalent to $15 per month required. Intro rate first three months only, then full price plan options available. Taxes and fees extra. See full terms at mintmobile.com. This episode of the Times Tech Podcast is sponsored by Vanta.
Let's talk about something that might be keeping you up at night. Cybersecurity. According to Vanta's latest State of Trust report, it's the number one concern for UK businesses. And that's where Vanta comes in. Whether you're a startup, growing fast or already established, Vanta can help you get ISO 27001 certified and more without the headaches.
And Vanta allows your company to centralize security workflows, complete questionnaires up to five times faster and proactively manage vendor risk to help your team not only get compliant, but stay compliant. So stop stressing over cybersecurity and start focusing on growing your business in 2025. Check out Vanta and let them handle the tough stuff. Head to vanta.com forward slash the times tech to learn more.
Because when it comes to your business, it's not just about keeping the lights on. It's about keeping everything secure. Before the AirPod arrived, I probably spent 7% of my life untangling my headphone cord that would be jumbled up in my pocket and then I would just have to undo it like some kind of puzzle.
And that's what I was doing before I got on here, which is why I was late, because I have these janky little headphones. But I feel like instead of spending 7% of my life untangling headphone cords, I probably spend 4% of my life looking for my AirPods. Oh, me too. So it's a marginal improvement. So what we're saying is Apple hasn't really solved the productivity puzzle. It's just created another. Yeah.
This is the Times Tech Podcast. I'm Danny Fortson. Danny, back in the valley. And I'm Katie Prescott in the city of London, still in London. What a week for tech. It's been and our podcast. Yeah, so...
Ahead of the Paris Tech Action Summit, OpenAI's boss, Sam Altman, sat down with us together in London earlier this week. Do go and listen to that episode if you haven't already. Yeah, the Action Summit, of course, not the Safety Summit. We are going to revisit that interview today to put it in the context of the AI Action Summit and in the context of what Elon Musk went and did
As soon as we put Sam Altman on the Eurostar to Paris. Actually, I don't know if he was on the Eurostar or not. I just made that up. Maybe he might have gone in a private jet. But yeah, so imagine the scene. It's late Monday morning.
Sam Altman comes in, we shake his hand, thank him for coming to our studio in London. We have the cameras there. Everybody's very excited. Took some glossy photos. I went off and wrote a couple of stories for the paper afterward. We edited the podcast, put it up out there for the world to see. And then of course headed to the pub for a very well-deserved pint of ale. I miss English ale. And it was a very good pub actually down a tiny alleyway in the city. Very, very London. Very London, cobbled street and lamps.
Then our phones started lighting up because there was a $97.4 billion question when Elon Musk said, Sam, OpenAI, let me take that burden off your shoulders for just under $100 billion. And Sam said, I'm good, but I'll buy Twitter for a tenth of that for just under $10 billion as a joke, obviously. Ouch.
Yeah, he can afford it. I mean, can he afford it? I mean, I know he's a billionaire, but either way, I think that it was very swiftly rejected. What do you reckon? Is this a serious play by Elon? No. I mean, it's got a snowball's chance because for a couple of reasons. One, last I checked, Elon Musk has seven jobs, eight jobs.
If you include his unpaid position within the government. And of course, he's running a direct competitor to OpenAI. And that competitor is far behind. So the question is, how would this even work? Because the whole idea was...
He's mad. Mad, you mean angry, not mad. Exactly. He's angry. Just translating for the British audience. You go back 10 years ago, he set up OpenAI with Sam Altman and they set it up as has been well documented as a nonprofit for the benefit of humanity. The whole idea being we're going to create artificial general intelligence, this thing that's better than the best human at anything possible.
But we're going to keep it open as the name implies. And this is going to be basically a public service. And then along the way, they part ways quite acrimoniously. And we can talk about that. And then this is his the latest kind of version of his attempt to get back at Sam to kind of retake control over open open AI, because of course, he sued him now three or four different times.
over Sam's conversion of the company from a non-profit to a for-profit company with a kind of non-profit on the side. I think we should get into the story of that relationship because it really is. And we talked about the broligarchy that's around the Trump camp the other day. It's like a medieval court. And this is like a medieval –
story in itself because as you say this was started as a non-profit really Musk reacting to what Google was doing at the time and it led to it I mean in the sort of theme of the medieval court it led to a massive falling out with Larry Page at the time of Google but he pulled together Sam Altman
Peter Thiel, and they were looking at this idea of promoting AI safety. Now, fast forward to 2018, Elon Musk tried to fold open AI into Tesla, and that was rejected, and that led to this great schism.
when Sam Altman started a for-profit arm. And it's very, very complicated when you look at the structure of OpenAI. And actually, it's a real life lesson in trying to sort these things out early and legally because part of the reason Sam Altman was pushed out of the business over that couple of days was about the company's structure and its aims. And I think now what he's trying to do is really create a very clear structure around the business. But Musk, harking back to when it was started,
has still got massive issues with that. And OpenAI has come out and published their emails. They're like, no, no, no, no, sir. This is what happens. Yeah, exactly. We're about to show the world our receipts. So they did that in March last year. This is why it's kind of so baffling in a way.
Is that they show the receipts, they show these kind of back and forth, these emails from going back 2017, 2018. And the context is they start this lab, they start actually making real progress. And they're like, oh, we think we might actually be able to pull this off. We might be able to build AGI. The problem is we're going to need like billions of dollars to do it.
And, you know, they're a nonprofit and they all agree, look, the only way to do this is to kind of turn this into a for-profit or to have a for-profit arm, et cetera. And so they have all these negotiations with Elon Musk, who at the time was the main funder of the company or sorry, the nonprofit. And his whole pitch was like, yeah, yeah, let's turn into a for-profit. I'm going to control it. I'm going to be CEO. And by the way, I'm also going to control the equity. In other words,
We're creating this thing for the benefit of humanity. But the only condition is I alone will be able to control it. I'm the king in this court. So enter Reid Hoffman, a rival king who offered at this point to bridge the gap to cover salaries and operations.
And so that was a sort of challenge, I guess, to Elon's crown at that point, because there was someone else with power and money who could come in and help out. And so, of course, they don't agree with Elon. They're saying, well, like, this is kind of contrary to our mission of like, hey, we're going to help the planet and the human race and keep this open. And for the benefit of all, don't feel so great about you being the one person that can control it. And so he leaves in a huff and says, basically, you're doomed to failure.
And then he goes off and does other things. And then to his surprise, Sam Altman actually manages to pull off a number of coups, including, of course, leading up to Chad GPT. And all of a sudden we're in this new world. And Elon Musk has since raised billions of dollars for a direct competitor, which surprise, surprise, is not a nonprofit, is not, hey, this is open source. We're going to keep, you know, we're doing this for the good of the world. It's just a direct competitor.
In Grok. And so you'd start, you step back and it starts to look just like a business feud, but it's still being dressed up as like, you need to return this to its roots as an open source for the good of humanity type project. And so what it really boils down to is like two very powerful men kind of jousting over control of this thing that if you believe Demis Hassabis, Satya Nadella, Sam Altman, Dari Amadei,
Is going to completely upend society, completely upend the economy and suck away trillions of dollars out of all these jobs that are going to funnel to probably a handful of big companies. In that context, you're like, yeah, this is just a very bitter business dispute. And it's one that thus far hasn't.
Musk has lost. We've been looking at the emails that OpenAI has published, but to give Musk his due, just to quote some stuff from the lawsuit that he's put out there,
Which plays to that Kingmaker point. There would have been no OpenAI without Mr. Musk's founding contributions and early leadership. He continued to make contributions to OpenAI from its founding right up until 2020. Then it accuses OpenAI really of breaching the founding agreement. The fact that he was critical and was the primary funder, that's all like inarguable. And you can talk about whether...
open ai has veered from their mission i think they have because they're a for-profit company now you know kind of in hawk to microsoft the largest company on the planet who are the biggest backer
but that's kind of almost beside the point of this conflict of like they're like look elon we want to continue working with you we just don't want you to have complete control over this foundational technology he goes off in a huff they go on to have success in a way that he doesn't really like and now he's trying to kind of throw his toys out of the pram and slow them down while he is building a competitor the only other thing i would say about the bid that elon musk did
Sam Altman has turned this nonprofit into a company that is raising money at a $300 billion valuation. Going from effectively zero with the help of Elon Musk at the beginning to now a $300 billion company, that is legendary in terms of just a business story and a CEO story. So the idea that all the investors in the board would ever be like, you know what?
You're right, Elon. We're just going to chuck that away and trust this whole caper to you, especially when you have seven other jobs. It just doesn't pass the smell test. I just feel like you're worried about him being overstretched. It just feels like you have seven jobs and one of them is like,
sleeping on in a sleeping bag on the floor of the white house as taylor swift said on her t-shirt a lot going on right now i've got a lot going on right now like maybe exactly i'm kind of like oh come on come on on the kingmaker point though on the power broking point it is interesting to see the figures who you think might who one thinks might line up behind sam altman on this so he's got a relationship with massa son from soft bank now satya nadela
is a backer, a very close backer. And you look at when he was fired from OpenAI, he got offered a job by Microsoft instantly. And so he's got, yes, Musk has got Trump's support and is in the Trump camp, but Sam Altman's got some incredibly powerful friends in Silicon Valley. Absolutely. I think, to be honest, this whole Musk thing, I mean, we're talking about it because it's the richest man on the planet trying to take over
arguably one of the most consequential companies on the planet, but I think it's ultimately a bit of a sideshow. He also stole our thunder, Danny. He did. It was a really kind of big punch that we didn't get to ask Goldman about. Exactly. I'm sure he's a devoted listener and he was like, you know what, let me try to steal their thunder.
I'm just going to throw this cat amongst the pigeons. Hours, hours after we published our pod. I don't think he is a listener, actually. I think we count as legacy media. That's actually true. Sorry. That's actually true. This episode of the Times Tech Podcast is sponsored by Vanta.
Let's talk about something that might be keeping you up at night. Cybersecurity. According to Vanta's latest State of Trust report, it's the number one concern for UK businesses. And that's where Vanta comes in. Whether you're a startup, growing fast or already established, Vanta can help you get ISO 27001 certified and more without the headaches.
And Vanta allows your company to centralize security workflows, complete questionnaires up to five times faster and proactively manage vendor risk to help your team not only get compliant, but stay compliant. So stop stressing over cybersecurity and start focusing on growing your business in 2025. Check out Vanta and let them handle the tough stuff. Head to vanta.com forward slash the times tech to learn more.
Because when it comes to your business, it's not just about keeping the lights on. It's about keeping everything secure. If you're a parent or share a fridge with someone, Instacart is about to make grocery shopping so much easier. Because with family carts, you can share a cart with your partner and each add the items you want. Since between the two of you, odds are you'll both remember everything you need.
And this way, you'll never have to eat milkless cereal again. So, minimize the stress of the weekly shop with Family Carts. Download the Instacart app and get delivery in as fast as 30 minutes. Plus, enjoy $0 delivery fees on your first three orders. Service fees apply for three orders in 14 days. Excludes restaurants.
Ryan Reynolds here from Mint Mobile. I don't know if you knew this, but anyone can get the same premium wireless for $15 a month plan that I've been enjoying. It's not just for celebrities. So do like I did and have one of your assistant's assistants switch you to Mint Mobile today. I'm
I'm told it's super easy to do at mintmobile.com slash switch. Upfront payment of $45 for three-month plan equivalent to $15 per month required. Intro rate first three months only, then full price plan options available. Taxes and fees extra. Feeful terms at mintmobile.com. But should we step back? Because I think the other thing that's just, you know, with the Paris summit come and gone, I think it's worth drawing out a few themes from there.
Because it was like the third international summit after Bletchley Park, which is all about safety. And then the one in Korea, which was just, I'm not really sure what it was about, but more kind of like AI, we need to regulate it, etc. And this one was called the Action Summit.
leaders from 100 countries. J.D. Vance was there, Emmanuel Macron, Chinese leadership, etc. And the whole idea is like, all right, we actually need to start taking action. We need to kind of come together as a world to figure out what we want this technology development to look like. What does it mean for society? What guardrails should we put on it? And stunningly, it was a damp squib. They came out with this
It's kind of a milquetoast statement that didn't say much. The UK didn't sign up to it. The US didn't sign up to it. J.D. Vance is like, don't overregulate this thing. You're going to kill it. It kind of gets to the point which we talked to Sam about. This feels like a race, like an arms race.
And we haven't got to the point where we're like, okay, can we at least agree some ground rules here? Everybody's like, I want to win. Well, Olman's talked about it in the past as being a bit like the early days of nuclear power. Yeah. And it is important that people talk about it. I actually thought it was quite strange that the UK didn't sign the agreement. And I couldn't quite work out why. Because why?
the UK has done in the past. It obviously kicked off all these summits by starting the Bletchley Park Summit in the first place. Yes, that was under a different administration. Yes, as you said, its focus was safety, which is
a bit different in this. I think this summit wrapped in many more issues, copyright being one of them, for example. But it did feel odd to me. And there were a couple of suggestions as to why the UK didn't sign it. Number one, of course, was that it was trying to count out to the US in some way. Number two, that AI is a massive industry in the UK, and there was some fear that it would
Curb development, although I find that strange because obviously China is another huge developer and did sign it. Then this idea that the restrictions on energy or commitments on energy use in AI were in some way not really practical or properly outlined. But it did seem weird.
Given that the UK wanted to see, or was trying to position itself certainly under the last government as a world leader in sorting out the regulation, but also sorting out the regulation around AI and saying, okay, we've got this new technology, but we are going to champion the various guardrails around it. Be that on safety, be that on copyright.
whatever and it didn't i don't know if if you've got thoughts on on the us angle or or why jd vance kind of swerved it yeah again i think it gets back to that arms race point i think i've done a fair amount of reporting on the the deep seek moment that we talked about a couple weeks ago and that really freaked out people in washington that really freaked out people in the pentagon
And if you get back to that idea around this being akin to nuclear weapons in certain ways, you know, if you can all of a sudden have a super powerful AI that's like powers thousand dollar drones and you can do a drone swarm and to go send them off to do something terrible, et cetera, like it's seen as what they call like the next offset. Like whoever controls this technology, whoever has AI supremacy, that is the top of the waterfall of global power. And so I think that deep seek moment,
Show that China was much closer than they had previously thought. And in this kind of great power moves, they're kind of like, well, we have to basically keep the shackles off our top US developers and make sure that this is safe.
where America remains the leader. And generally, there's a kind of allergic reaction, especially from Republicans and on the right to kind of regulation generally. But the shackles are still on in Europe. And we asked Sam about that. And he said that would probably slow down the releases of different models in Europe. Let's say the European rules require that it takes...
three months of testing for a new model or a new modality or something like that. Let's say five months. And let's say our improvement cycle is we have a new model come out every three months. We never catch up. We never get to deploy. It's always something that's like pretty far out of date. And again, that might be, I'm not here to say that's like the right or wrong choice. I obviously have a personal bias, but I think people can reasonably disagree on that. But I do think you have to look at the whole package here. And if we are right, that this technology is going to be so impactful to not
And changing every few months to not get the, not just the economic benefit of that, but the sort of like influence on the world that comes from mastering those tools and integrating them into an economy.
That comes with a cost. I thought the US reaction was interesting because I think in terms of DeepSeek, it also was a wake-up call that maybe the curbs on chip exports to China aren't working in the way that they were designed to, which was to try and hold back advances in technology. We don't really know how...
the DeepSeq model was developed. We know what we've been told. When we spoke to Professor Neil Lawrence, he was saying some of the engineering looks extraordinary, but there's also the possibility that they've been using chips perhaps even offshore.
to train the model. Yes. Yeah, exactly. You know, and that I think that's also presumably scaring. There's a lot of talk about Singapore. Like maybe like NVIDIA is selling a lot of their best chips to Singapore and like Singapore is like a single city state. Yeah, exactly. So perhaps these exports aren't working. There have been hints of that in the past. Bloomberg's done some brilliant work looking inside Huawei phones and looking at the chips that have been used there and where they've come from and how advanced they are.
but certainly the deep seat model was a big, a big wake up call there. Yeah. This, these models are getting so powerful and so good. For example, deep research, which we talked about, which is this, this reasoning model that you can kind of send off to do more complex tasks. It'll kind of go do its work, check its work, generate reports, you know,
And there's this guy, Tyler Cowen, he's kind of public intellectual, really well respected out here. He's an economist.
And he said, he put out this post last week and he was like, hey, I was messing around with opening eyes deep research. And this is what he said. He said, I had to write a number of 10-page papers for me, each of them outstanding. I think the quality is comparable to having a good PhD-level research assistant and sending that person away with a task for a week or two or maybe more, except deep research does the work in five or six minutes. Wow.
I side there because, as you know, I'm writing a book at the moment and I think it's an amazing idea, but I wouldn't yet trust it. You know, would you to write an article you're going to publish? I mean, it sounds amazing, but I still...
feel like I'd want to check everything that came back myself. Yeah. And I think that's what's going to be really interesting over this next year. There's the credibility gap is still like for you, for me, for a lot of people, I think it's still quite wide. But I think for a lot of people, it's there is no credibility gap. They just are going to believe what an AI says. And again, that's why I think we're entering into really funky territory. The BBC did some great research this week where it looked at all of the different AI chatbots and it found that
Four of the biggest ones are inaccurately summarizing news stories, which is quite amazing. 51% of all AI answers to questions about the news were judged to have some significant issues.
20% of AI answers which cited BBC content had factual errors. So it's really cool. But if you're a researcher, if you're an academic, if you're an author, if you're a journalist, you'd hold yourself back from using it. And I think that was what was one of the most interesting things for me listening to Sam Altman when we interviewed him, was he talked about AGI coming so quickly. Just because it's top of mind, this deep research thing that we just launched. My...
This is not a scientifically rigorous thing. My vibes-based estimate is that does about 5% of all tasks in the economy today. One feature, 5% of all the tasks in the economy today. 5% of economic work's being done.
But in the real world, I don't see it being used yet. And I still feel there's this gap between what they're developing and its real world application. And that comes from the trust point. But I think it also comes from people's knowledge and understanding of it, which is still very low.
And I know it's this like hot topic in Silicon Valley. Yes, they're discussing it in France, but it's still very much up here. And I don't feel that it's percolating in the real world economy. By the way, vibes based estimate is a thing I'm just going to use now. Oh, yeah. Absolutely. Like if my wife asks like when's dinner going to be ready? I'm like, I don't know.
vibes-based estimate is sometime within the next 24 hours. I love that you used that as an example, not when are you going to be home, which is what a lot of men would say. When are you going to be back from the pub? I'm cooking and cleaning and doing everything. When's my dinner on the table? Honey, I can only give you a vibes-based estimate. My vibes are bad right now. I'm tired. It's another way of saying I think. It is, actually.
I'm going to tell my kids to give me vibes-based estimates. That's great. What was the thing that struck you most about sitting down with Sam Alton? It's just the idea that there's this kind of gauzy optimism of like, you know, look, this is going to be...
Like what Sam said, he's like, we're not going to all, he's like, I just don't believe in a future where we're all just going to have nothing to do. And we're going to just like do drugs and play video games. Although some people might enjoy that. But he did say, you know, this is going to, and everybody seems to agree, it's going to be painful. It's going to be messy. There's going to be a lot of jobs that just disappear. You know, that messy middle is kind of hurtling toward us. And nobody has any good answers right now about like, yeah, there's this like,
It's like we're at point A and point C is like this utopian future where we cure cancer and Alzheimer's and we live forever and we have clean power and it's all done by AI. Humanity's last invention has saved us. But between point A and point C is this point B. What's that look like? Yeah, I think that's absolutely right. And the disconnect as well between the kingmakers in Silicon Valley, between this medieval court, this...
wild sequence of battles between the kings in Silicon Valley. And what real people, what the average person really understands about AI or is even experiencing today is still absolutely huge. And I think it's easy to forget that because we're covering it day in, day out, and we're following all the twists and turns of it to just think in the real world, this hasn't really sunk in yet. Was that your main takeaway?
I just enjoyed getting to spend some time with him. As you have, OpenAI is a company that I've covered really, really, really closely. And so it was very, very cool on a personal level to get to see what he was like. I found him, you know, he's clearly extraordinarily intelligent and it's interesting to see some of the way that he jibed around our questions and reacted to what we were saying.
I found him very measured, very, very diplomatic. It's extraordinary how young he is, you know, and his life he has to deal with, with the power that he has now is completely crazy. And I got the impression he still hadn't wrapped his head around that. And that he quite likes it.
Yeah, probably. Well, look, I think you should go and get some rest. I do too. I feel like you've worked very hard this week. It's not an easy turnaround. Do go and listen to that episode. Totally worth it. Very fun. And it was fun to be in the same place with you as a carbon-based life form as opposed to, you know, a square on my computer. What did you call us last week? Oh, Autonomous Data Production Units. That thing. Oh, goodness. It's all going to be fine. It's all going to be fine.
Another fabulous week. Two pods. You're welcome, dear listeners. And before we go, please send any of your questions, your thoughts, even your criticisms, if they're gentle, to techpod at thetimes.co.uk, techpod at thetimes.co.uk. This episode of the Times Tech Podcast is sponsored by Vanta.
Let's talk about something that might be keeping you up at night. Cybersecurity. According to Vanta's latest State of Trust report, it's the number one concern for UK businesses. And that's where Vanta comes in. Whether you're a startup, growing fast or already established, Vanta can help you get ISO 27001 certified and more without the headaches.
And Vanta allows your company to centralize security workflows, complete questionnaires up to five times faster and proactively manage vendor risk to help your team not only get compliant, but stay compliant. So stop stressing over cybersecurity and start focusing on growing your business in 2025. Check out Vanta and let them handle the tough stuff. Head to vanta.com forward slash the times tech to learn more.
Because when it comes to your business, it's not just about keeping the lights on. It's about keeping everything secure. What makes a great pair of glasses? At Warby Parker, it's all the invisible extras without the extra cost. Their designer quality frames start at $95, including prescription lenses, plus scratch-resistant, smudge-resistant, and anti-reflective coatings, and UV protection, and free adjustments for life. To
To find your next pair of glasses, sunglasses, or contact lenses, or to find the Warby Parker store nearest you, head over to warbyparker.com. That's warbyparker.com.