OpenAI’s Crazy Week, Meta’s New Orion AR Glasses, MKBHD’s Wallpaper App

0:00OpenAI is in chaos as leaders leave, Sam-eyes equity, and leaks abound. Meanwhile, I try Meta's new Orion augmented reality glasses, and YouTuber Marcus Brownlee, also known as MKBHD, launches a poorly received wallpaper app. All that and more is coming up right after this. Welcome to Big Technology Podcast Friday edition, where we break down the news in our traditional cool-headed and nuanced format. Here, as always, on Fridays with Ranjan Roy. Ranjan, I have never been more excited to record an episode than I am today. This week, we have been texting back and forth and back and forth because there's been so much happening.

0:39So I've been waiting for this all week, Alex. We have, I mean, the drama continues to unfold at OpenAI and big changes for the corporate structure look like they are coming up. I also, as I mentioned in the intro, got a chance to head down to Meta's Menlo Park headquarters to try their Orion augmented reality glasses, which we'll talk about. and honestly this story about Marcus Brownlee and his wallpaper app is just like one of the most fascinating things I've seen. Just kind of like having a reviewer ship his own product and then all the people come back and say hey it's not that easy is it. So we have a packed schedule this week.

1:14Let's just start with the OpenAI news. There's so much that's happened but to me the big story of the week is that OpenAI is moving to basically sideline the non-profit board and become a traditional business. This is from Reuters. ChatGPT maker OpenAI is working on a plan to restructure its core business into a for-profit benefit corporation that will no longer be controlled by its non-profit board. The non-profit will continue to own a minority stake in the for-profit company, and the move could also have implications for how the company manages AI risks. Now, this is taking place in the context of OpenAI, raising about$7 billion at$150 billion valuation.

1:57So with that big money coming in and moving to this public benefit for-profit company comes, of course, the important issue of how Sam Altman, who for the entire time has said he's doing this as a benefit to society and he has enough money. Well, he's going to be rewarded. There were some reports out there this week saying he's going to get 7 % of the company or$10 billion. He's since denied it, but that doesn't change the fact that Sam is going to have a lot of money on the line here, and OpenAI is finally moving towards a more normal structure. What did you think about when you heard the news, Ranjan?

2:33I was actually kind of happy about this. And I think normally, amidst all this drama, this might have been the kind of thing that I would be questioning and a bit skeptical about, but this needs to happen. And I'm at least glad people are talking about it, because for probably a couple of years now, every time we talk about open AI, we talk about how weird and odd this entire capital structure is. And to actually just recognize this is a for-profit business. And this is a business that is making money, is also losing money and needs to make a lot more money to just make things clean and simple and give Sam equity and make it a for-profit company and people investing billions of dollars, like actually have a potential financial return that's clear.

3:20I think this is a good thing. I'm with you 100%. I saw the news. People were reacting poorly to this news. Of course, it does make Sam look a little bit hypocritical. We'll get into his financial package in a bit. But it does also solve a lot of the issues, or not completely solve, but start to address a lot of these issues that we've brought up over the past year. And really since last fall, there's been this glaring problem with OpenAI, which is that it has this nonprofit board, but it's got billions of dollars invested in it. And it just made, like you said, the company strange from so many ways and led to the problem of Altman's ouster and back in his return.

3:58And it was clearly operating as a for-profit company with this quote unquote nonprofit structure. It never made sense. And this will make the company at least act a lot more in line with a normal business. But if you think about this, this is open AI going from what was established as a nonprofit company that was going to do research for the public good to a for-profit standard style company, maybe a public benefit company, but still a standard as you get. And there's trade-offs here. I mean, the pro is that you're less weird, you get a lot of money and you can do the work because the work in AI right now needs that type of money.

4:42Like when Elon Musk and Sam Altman started the company with$50 million from Musk, just didn't need as much money. But now, you know, when you have training runs that cost multi-billion dollars to complete, you need the cash. And so they're able to do the work by being a company like this. But then there's the negative, right? And the negative is that you work for investors and you're no longer sort of aligned as much as you can be with, I don't know, the public good. Now you're simply working for the bottom line. What do you think about those trade-offs, Ranjan? Well, I think that idea that now you're working for investors is at the heart of this.

5:19And we talked a lot about this last week, and we'll definitely get more into the leadership changes, but that tension at the core of, is this a research house that's trying to create a safe AGI that's going to save humanity? Or is this a business that takes investor money and tries to see a return and delivers and ships products and has users? And I think it's good that if at least at the high level, the expectations are set very properly, we'll dig more into whether that actually will or could work. But again, I think at some point this has to happen. I'm more curious, though, about the actual mechanics of it.

5:58And I mean, I was trying to read through all the reporting, and it's still a bit unclear around what happens to Microsoft's investment, what happens to the capped for profit model, where it was the cap was set at 100 times the initial investment. So if an investor put in$10 million, the potential returns were capped at$1 billion, which obviously sounds dramatic and amazing. And everyone would be happy to get 100x return. But I think overall, there's been so much complexity in the initial setup, how they actually unwind that into a straightforward for-profit company. It's still pretty unclear right now.

6:40Yeah, that's going to be fascinating to watch play out. And can I just say, I'm a little bit sad that in some ways, this episode shows that you simply cannot work outside the system, that there is a system and you must work in it. And if you decide that you want to try something different, you can't. And I understand there's a lot of good that's come from the system. But it also sort of saddens me a little bit that the other way was not possible. Is that irrational? I think that's highly irrational. As a tried and true capitalist, I mean, I believe the capital markets have evolved and America is a great place to build a business and invest in because of all the structures and rules and systems that are in place.

7:28And I think that's a good thing. And I think that's why all these things work the way that they do. And I think the way they tried to approach this, if it was not actually trying to drive huge financial returns. And they didn't need to raise billions and billions of dollars at a high valuation. They could have taken donations. I don't know. I mean, it could have gone in any other direction if you wanted to create this truly hybrid structure. But the fact that they approach this like any other VC-backed business, I think it should be a sign that the way VC structures work can be a good thing. The way capital markets work is a good thing.

8:09And I think that playing by those rules is a good thing. Okay. Well, look, even if you're right, it isn't going to be clean and it isn't going to be neat as the company makes this transformation because it did begin as this research house. It has that research house DNA. There are a lot of researchers who were attracted to that. And now as it moves towards more of a business in a traditional sense as a company that's going to be worth more than 150 billion dollars if this round goes the way that it's hoping there's going to be some disgruntlement there's going to be some trade-offs who's going to get paid how are they going to get paid who's going to have the power what are they going to work on you know these are big questions and so we saw a lot of this tension come out in the fore this week that's why we're calling it open ai's mad week because first of all the chief technology officer mira marati she's she decided that she's leaving and announced her exit this week.

9:04If you remember, she was the person that took off, took over for Sam Altman when he was initially ousted by the board. So she's leaving. Remember, Ilya Sutskever is gone. Jan Schulman, who was another co-founder, is gone. He's at Anthropic. Jan Laika, who ran Super Alignment there, he's gone. He's also at Anthropic. Greg Brockman, the president, he's on leave. So Maradi, she's out. She announces her exit this week. And then almost immediately afterward. And the timing was very interesting because it was so abrupt. Someone named Barrett Sov, who's the VP of research and post-training at OpenAI also said that they were leaving.

9:43Right now, Barrett writes, feels like a natural point for me to explore opportunities outside of OpenAI. This is a personal decision based on how I wanted to evolve the next phase of my career. Very interesting. And then, of course, Bob McGrew. Okay, Bob McGrew, we talked about him two weeks ago when OpenAI released 01, that reasoning model. And he's the person that led research there. And he's also leaving. By the way, Bob McGrew is responsible for what I think might be my favorite quote ever coming out of OpenAI. And this is reported in the information. He had a meeting with colleagues after he said he was going to leave this week.

10:25He's been at OpenAI for eight years, and he told them he was feeling tired and he had a pool he hadn't had a time to swim in. He's going to go swim in that pool. But it is just another massive exodus coming from the top of OpenAI. And this is from the Wall Street Journal story reporting on it that Marati is one of more than 20 OpenAI researchers and executive who quit this year, including several co-founders. Is this just, you know, normal course of business? It seems like there's a lot of exits. I mean, even Sam Altman, when he addressed it, was like, yeah, we are going to just handle this the way that we handle the other high profile exits.

11:02And it's like, everything's OK. But why are you having so many? Well, we are a Bob McGrew fan club. So let's just recognize that his departure as head of research. We'll bring the beers. Let us come to the pool. We're down to swim. We're ready to take that swim, Bob. But it was funny because the Barrett Zoff quote, who's a VP of research, normally when you see any kind of executive departure quote, especially when people, you know, I'm leaving to spend more time with my family, it always feels a bit contrived. But when he said right now it feels like a natural point to explore new opportunities outside, I actually believed it.

11:43I really believed it because, again, we talked about this a lot last week, that tension between researchers and kind of bloodthirsty capitalist business people or just simple business people and business-driven people. And we are seeing this play out in real time. And it's going to be really interesting. And I am, and we've talked about this a lot, I'm generally bearish on open AI and especially every development like this makes me feel more bearish. But in a way, this is a moment that maybe Sam Altman can reset the company. Maybe they do have the technology that's good enough to drive the productization and the monetization that needs to happen to actually realize these valuations.

12:30Like maybe he goes and hires a bunch of Oracle salespeople and, you know, old school software salespeople, and they actually move this thing forward. But it had to happen at some point. So if this is that moment, it at least gives them a shot. I think if they continued on the trajectory they were, this would have just gone down and down and down. And let me just go back to like what I was saying about how I'm kind of sad about how this is the only way to do it. because you start to hear some of the reporting that comes out of the company this week in particular about this transition to a for-profit and look i understand the way the world works i'm not naive but just listen to what's happening there this is from the wall street journal current and former employees say open ai has rushed product announcements and safety testing and lost its lead over rival ai developers they say altman has largely been detached from the day-to-day, a characterization the company disputes as he has flown around the globe promoting AI and his plans to raise huge sums of money to build chips and data centers for AI to work.

13:32So like you hear the dispute from OpenAI and then a couple paragraphs later, Altman has been in Turin, Italy at this week's Italian Tech Week. Yeah, but no, really the main issue here for me is just that like there are these rushed product announcements. I'll just give one example. This is, again, from the journal story. I'm just going to read straight. This spring, tensions flared up internally over the development of a new AI model called GPT-40, which we know about that would power ChatGPT and business products. Researchers were asked to do more comprehensive safety testing than initially planned, but given only nine days to do it.

14:11After the model launched, people familiar with the projects said a subsequent analysis found the model exceeded OpenAI's internal standards for persuasion. By the way, it's amazing they measure this. What is persuasion? It's defined as the ability to create content that can persuade people to change their beliefs and engage in potentially dangerous or illegal behavior. The team flagged the problem to senior executives and worked on a fix, but some employees were frustrated by the process, saying that if the company had taken more time for safety testing, they could have addressed the problem before it got to users.

14:45Now listen, I'll be the first to tell you that I think some of these alarmist, doomer concerns about ai have been overblown but that being said if being a normal quote-unquote company and being a for-profit company has people shipping things that is scoring poorly on a test an internal test about persuade that that find them finds them able to persuade people to do dangerous or illegal behavior that is an emoji red flag for me i'm putting it up three emoji red flags yeah but i mean google told us to eat rocks let's not forget i think all of the generative ai test like these products being shipped we're coming more and more into the understanding that there's bumpy roads ahead for everybody and that's just going to be part of it i think to me so there's two parts of the rushed product releases there's the safety element but then there's also the actual products themselves.

15:39And to me, my emoji red flag has been OpenAI. I think what they did better than anyone else was actually let the products tell the story. The magic of ChatGPT wasn't, I mean, at that point, no one even knew what generative AI was or cared. They just saw that this thing did things that were unimaginable. And more and more, like over the first, you know, call it 2023, the beauty of OpenAI actually shipped things while everyone else gave us demos. Meanwhile, they are now moving towards that model. Sora, I think was debuted in January or February, still has not shipped. And there's reporting that they're still working to actually improve the model, which to deliver higher quality, which makes me think that those demos that we were shown were not actually for real and were just demos.

16:31and also that it takes a long time there was there's reporting that it took like 10 minutes to generate like a 30 second video or something like that which is just i think too long unusable exactly like unusable at any kind of consumer level but even the way they position how greedy are we by the way we're so greedy it's like make us a magic ai video but take one minute not 10 minutes to do it but it's just no i don't think it's greedy well because i don't think it's greedy because Sam Altman, there's even this Twitter exchange where I think someone was asking about multimodal voice and he had said something to the effect of, can you please take a minute to thank us for like the magical abilities we've given you?

17:10I still, and I've been using their multimodal voice and it is magical and it's amazing. And that is letting the product tell the story. But even the way they positioned O1 as reasoning felt like they were just trying to, for their fundraising purposes, latch onto the whole agentic AI thing. Reasoning is the next big thing and trying to just push and shove that into the product when it's not that at all at the moment. It just felt like, you know, maybe an incremental advance, but nothing revolutionary. And now it feels like they are playing, I don't want to call out Apple again, but what Apple is doing with Apple intelligence, what Google's done a bit like demo first, that's dramatic, exciting to try to drum up excitement.

17:55whereas they were always the ones who actually gave us the product to use. Well, this is what happens when you're in the system and you play the game that everybody has to play. You sort of have to puff your chest up and make things up like search GPT and then hope your fundraise will give you a higher valuation. So this week, we're anti-capitalism. Is that what you're saying, Alex? No, I'm also pro-capitalism. I'm just pointing out that like, it's so interesting how like going full capitalism incentivizes all these behaviors. There's no happy, there's no, I think the way to think about capitalism is it's the, it's the, um, what is the least worst or the, the best bad system?

18:35I don't know how, how people talk about it. I think that's true. I'm a fan. Obviously wouldn't want to do anything else, but like, yeah, it is to me like a little bit, just like, oh, you, you flip the for-profit company switch on. And then all of a sudden you're like doing vaporware demos or not i don't know if you know demos where we haven't seen the product you're making up you know these new business lines because you need the money like it's different i'm just highlighting the trade-off no no i i agree but to me it's then don't raise billions of dollars and yes on one hand but you need the billions to operate if you're going to play at all in ai you need the billions because it costs so much okay i'll give you it's it is a weird tough balance and I'm I am always on the side of at least some kind of competition for the big tech companies and the rise of open AI I've been hopeful around because it actually was probably the first time we saw any fear in the last decade from the big tech companies but it's a tough one I guess do you think this is a bad week or a good week for open AI I mean short term it's it's bad but long-term it's necessary and will be good.

19:47Yeah. Okay. I will agree. I will go with, I think that's a good week. I think if it is, I just said, if there is a chance that they're able to regroup restructure and do what they need to do, all of this has to happen. So the faster it does, the better. Right. All right. Let's talk quickly about Altman's compensation. So there were some reports that he was going to take 7 % stake in the new open ai but on thursday he said that that was ludicrous and he said that investors this is according to information we're pushing for an equity grant to align his financial interests with those of open ai and look like people have been criticizing this all week because sam has said like quite publicly that like he doesn't want to get paid for this and why can't people get it through their small heads that he has enough money but again like thinking about like short-term weirdness, long-term benefit.

20:40Yes, you want to align his financial interests with open AI. That to me just seems like it makes a ton of sense. Yeah, exactly. The compensation, the equity stake is probably the most central part of this whole moment in conversation because exactly the idea that he was ever doing this for no money at all was ludicrous. So the idea that we can all be honest about it and that he has. And 7 % sounds nice as a founder diluted after a few big rounds of funding, seems about correct. And just that we move in that direction and just go there. But that's even why rather than just simply owning up to it, denying it at an all hands meeting, and then us still having this back and forth, I guess that would be my next emoji red flag that they're still not just going all in and just moving on this, that they're actually still trying to have it both ways.

21:40Do you know what I view this whole chapter as? It's kind of like Elon Musk's revenge or Elon Musk's time bomb, because when he set up the company initially with Sam, he was like, basically, it's going to be a nonprofit and it's going to be research focused. And if it starts to go in any different direction, there will be this sort of time bomb where this very painful moment will take place. And it looks like that's what's happening. I would give Peter Thiel that kind of credit to pull off something like that after the great Gawker Hulk Hogan stories of the past. I think if anyone could pull off that long game, carefully, meticulously planned and plotted, I think I would definitely think Peter Thiel could do that.

22:21I don't know. I was thinking more like if this thing, basically it was inevitable that if this thing does end up going to like the more for-profit side, it was going to go through issues sort of baked into the founding agreement of the organization. And here it is. Maybe I'm getting into this a little bit. If Elon Musk is listening. Elon Musk's revenge, let me tell you. If Elon's listening right now, I can see him tweeting this, that that was the plan all along. And that would actually make him look pretty good. Yeah. And let's just take one moment on Sam, because like there's been a lot of discussion this week about how OpenAI is cooked without, you know, me or Marati and the rest of the research leaders and with all these executives leaving.

23:10And I would like to say that that seems outlandish to me. He's going to raise this money. He still has a tremendous amount of talent. There's more clarity on the direction they're going, like we've talked about. it. And remember, I mean, you know, think about all the quote unquote visionaries in tech. They all knew that the transformer model was out there. They all could have done this. They could have done it at DeepMind and they didn't do it. And instead, I think under Altman, this company has definitely shown that it has direction. It has product chops. And it was the one that started this whole thing, the generative AI moment.

23:46So I'm not counting them out. I'm not counting them out either i think especially because i mean i've argued this that to me even though he is still saying the next big generation of models and agi is where it's at to me this battle is won on being able to productize what you already have and he has driven that better than anyone else and chat gpt and a lot of what they've released actually is the best product and that's what they had been winning on or have been winning on. So yeah, I don't think at all researchers leaving at this stage with all the initial research that has been done and that's already done.

24:26I agree on that. I don't think I'm not counting him out at all. Oh, and just before we went live, we got some news from CNBC. Kate Rooney reported this, that OpenAI is telling investors that this round is going to close next week. So Sarah Fryer, the CFO, said the funding was oversubscribed and would close by next week. And she says, even though there have been these departures, collectively, we remain laser focused on bringing AI to everyone and building sustainable revenue models that fuel our operations and deliver value to investors and employees. And yes, she's been doling out a lot of money to researchers who've like gone and gotten offers and come back and they have to keep them keep them there.

25:08So maybe luckily for the company, you know, it's going to close this round, it's going to restructure. And then maybe I wouldn't say all this drama, there's no way that all the drama is behind it, but maybe this chapter of drama is behind it. Oh, wait, I had totally forgotten Sarah Fryer had become CFO. I think, yeah, it was probably this summer or maybe in May confirming it. That is interesting to me now thinking through what's happened in the last few months because again she is long-term high-level traditional tech executive ceo of next door cfo of square so that is interesting to me now thinking about that that like you know the the traditional tech talent has been brought in over the last few months and we're seeing the ramifications of that we're seeing it's becoming a good old-fashioned corporation okay let's uh close this out well actually maybe we have one more thing to talk about but first uh sarah fryer knocks on your door and says ron john i want to offer you a position in this funding round give us give us a thousand dollars you'll get the same terms as the other investors uh it's worth you know we're going at 150 billion dollar valuation you giving her the money wait it's not a billion dollars she's asking me for alex no this is a hypothetical of course i mean I would, I'm still a no.

26:30I still think I'm calling it a good week. I'm saying, but I still think the probability of success at this stage, at this valuation is a very, very low prospect, still overall bearish. What about you? Yeah, I'm giving the money. I don't know. Maybe it's Alex is in Alex. I'm in. I'm in. Um, I can't tell you why. I obviously think it would be a, it's a risky proposition, but you do have a high potential upside, especially if this stuff works out. So I'm in. I mean, I guess that's the thinking for the investors that are there. Although like, I wouldn't be surprised at all if I flushed that cash down the toilet.

27:09But one interesting thing that's also happening is the mainstream media sort of catching up to some of our skepticism and some of the people that we've highlighted on the show. Why don't tell us about that. Yeah, the New York Times had a piece this week covering the Goldman report that we called the canary in the coal mine, coal mine almost two months ago, I think, that really kind of hit hard on generative AI and the economic prospects. And I actually think this is kind of reaffirming what we were saying, because this is the kind of thing that Jim Cavallo, the Goldman Sachs, head of global equity research, who was kind of the central figure in this report that just said there's actually it was amazing that they had said that there's no real future, but still to invest in Gen A infrastructure because people are still going to be spending money on it in the near term because everyone is that's what everyone's doing no matter what.

28:07But I think this is actually a big moment because when The New York Times is now kind of picking up on this and making it a headline, will AI be a bust, a Wall Street skeptic rings the alarm. This is moving mainstream. And I think there's going to be more and more and more questioning of the economics. And I think that just kind of both affirmed it, but also means it's going to become more mainstream of a topic. So AI is something that has a huge amount of buzz, but may turn out to be nothing. Well, maybe not nothing, but not quite what we imagined. Not nothing. Can we go back in time and look at something that now seems like it might turn out to be nothing but actually could be quite big?

28:49And what I am talking about is the metaverse. Because this week, I was down in Silicon Valley and got a chance to try Meta's new augmented reality glasses called Orion. And let me tell you, these are a real technological marvel. You put them on, they do not feel heavy. You get 70 degree field of view. I use them to watch YouTube, make video calls, play games, scroll through messages. These things are pretty incredible. And all of a sudden, they put the larger dream of layering computing on top of the real world. And having that in a form factor that makes sense. They make it feasible. They put it into a product that makes you see mainstream augmented reality, mixed reality, for the first time.

29:45Were you watching when Zuckerberg made these announcements or have you read the coverage? And what do you think about these things? I am very excited about this. I think the idea that the metaverse could still live on, the metaverse dream of Mark Zuckerberg, is kind of amazing. And augmented reality, I've been very, very excited about for a long time. So this really becomes that kind of next, like the actual mainstreaming and kind of real innovation around it. That is amazing. um i mean i have to ask like was it i tried the magic leap in pretty early on when it came out and it i put it up there in like maybe top 10 technological moments i've ever had of just kind of pure and from in terms of gadgets like just painting around me and seeing all this stuff and but that still had a huge battery pack and was heavy and didn't work that quickly so the idea that and i mean was it that amazing because at least the way they talk about like the miniaturization of computing this is like one of the greatest advances obviously this is them saying that but it seems like other people said it really is a technological marvel is it that cool so this is going to be experienced when you use it it's going to be an experience that you're going to be familiar with like unlike chat gpt which came out of nowhere for a lot of of people, using these Orion glasses is going to be something people know because they've tried the Oculus or they've tried the Vision Pro.