Microsoft & OpenAI Visit Davos, Is The Vision Pro Dead On Arrival, Google Makes More Cuts

0:00Let's recap and react to the big AI conversations at Davos. Plenty of news items this week also have us asking whether the Vision Pro is dead on arrival. Google is making more cuts. And Meta has bought a boatload of NVIDIA chips. All that and more coming up right after this.

0:22You're used to hearing my voice on the world bringing you interviews from around the globe. And you hear me reporting environment and climate news. I'm Carolyn Buehler. And I'm Marco Werman. We're now with you hosting the world together. More global journalism with a fresh new sound. Listen to the world on your local public radio station and wherever you find your podcasts.

0:49Welcome to Big Technology Podcast Friday edition, where we break down the news in our traditional cool-headed and nuanced format. We have so much to talk about this week, including plenty of updates from Davos, where it seems like everybody gathered, And then real product news from Apple and corporate news from Google and Facebook, just buying up every NVIDIA chip in the world. Joining us as always here on Fridays is Ranjan Roy, who writes margins on Substack. Ranjan, welcome. Alex, are you in Brooklyn or are you in Davos right now? Honestly, the best thing about YouTube is that you can literally sit in on every single conversation from Davos and you don't need to go.

1:30And I've been there. All right, Alan, I have to drop it. I've been to Davos. This is where Alex is going to say I've been to Davos. I, Ranjan, have not been to Davos, but tell us about Davos, Alex. Having been there once, I was very, very happy to watch everything on YouTube and then actually be able to speak with people about what it all meant. And I think we're going to have a good time going through what happened there here this week. So we're going to do a new thing here. We're going to play a few recordings of some of the big names that actually spoke there and react to what they said and kind of unpack it.

2:06So the two speakers that I picked out were Satya Nadella and Sam Altman, not surprisingly. They've been driving a lot of the news in the tech world, in the AI world recently. And they spoke about some really interesting things, including the state of their partnership and where the research is going. And this is from their conversations with Brad Stone at the Bloomberg House. So, Ranjan, what do you say? Let's get right into it. Let's get into it. Okay, so first of all, in case you're wondering about how Microsoft feels about the OpenAI partnership, here is what Satya Nadella had to say about it.

2:42I feel very, very good about the construct we have. I feel at the same time very capable of controlling our own destiny. So it's not like that we are single threaded even today on Azure. And this is not about even open AI. It's all about reflection of what our customers want. Every customer who comes to Azure, for example, in fact, our own products. It's not about one model. We care deeply about having the best frontier model, which happens to be, for example, today, GPT-4. But we also have mixed raw as a model as a service inside of Azure. We use Lama in places. We have Pi, which is the best SLM from Microsoft.

3:23Now, doesn't that sound like Satya Nadella taking pains to really push home the point that they're not reliant on open AI? And doesn't it sound like Microsoft is trying to hedge away from open AI? I don't think he's taking pains. I think he's pretty confident about saying that they're not dependent on open AI. I mean, very clearly saying, you know, right now the best model is GPT-4, but we are, you know, using Llama and everything else, I think is pretty interesting. And again, on that, I've been using Microsoft Copilot recently more. Again, their public facing chat interface, because it actually gives you access to GPT-4 and Dolly 3 for free.

4:09You just have to be signed in. unlike chat gpt where you need to pay for chat gpt plus to get access to the latest models on a reliable basis and it's crazy to me because it they're giving it away for free and we've talked about it previously on the show but but it's clear to me microsoft is not going to be reliant on open ai what that partnership looks like going forward i i really don't understand what it is right now I don't know. What do you think Satya's vision is right now for open AI? Because he didn't seem to give a clear. That is a great question. I mean, I think that's one of the big questions that we're going to have to answer this year.

4:48I don't have an answer to it, honestly. And there was a conversation that was had on CNBC throughout the week about how impactful open AI was and AI in general was to Microsoft's valuation. Microsoft, we should note, just surpassed Apple as the most valuable company in the world again. And there were rumblings that OpenAI and AI itself has added a trillion dollars to Microsoft's valuation and market cap. And so what are you left with if they now are trying to distance themselves from OpenAI? I guess what Satya Nadella is really doing right here is he's saying, I'm riding the horse that I have, which is OpenAI.

5:26But I also know that there's going to be some sort of commoditization of the technology that they're building. assuming open AI doesn't continue to outpace the field and it will be difficult to do that. And so therefore, he's waiting on the hedge. But then the question is, does the hedge provide the competitive advantage to Microsoft if everybody is starting to be able to take advantage of that? Like one of the models he mentioned was Llama, which is open source. So where does Microsoft's advantage lie there? Is it productizing this AI? Maybe. I mean, Enterprise is definitely a place where this is going to factor, but it is very confusing.

6:02Yeah, I think especially a company like Microsoft, it's worth, there's multiple layers. When we say enterprise, we even have to think about it in two levels. There's a consumer software or enterprise software layer. Again, when I open Microsoft Word or PowerPoint, what capabilities do I have? And Microsoft's moving pretty fast in this. like, you know, they're way ahead of Google in terms of what they've actually launched and released to their customers. And then, of course, there's the more deep integrations within Azure, people building their own products. And in both of those, it's clear that they're making big strides and they're not dependent on open AI models.

6:43I mean, I think it's clear that the product level is going to potentially be a big competitive advantage for them. And again, I've been using Microsoft CoPilot a lot more. And just for fun, when Alex, when we were going over this week's topics, and I was like, I'll do this nowadays more when I'm looking up a news story. I'll check ChatGPT, BARD, CoPilot, see, you know, can you give me a list of articles around a specific topic? In great moments in BARD, Google's chatbot product, I asked, you know, can you list some news articles and provide a recap of what Satya Nadella and Sam Altman said at Davos.

7:26And the answer is, elections are a complex topic with fast-changing information. To make sure you have the latest and most accurate information, try Google search. Okay. So I really don't feel so bad for, I just bashed Bard on air on CNBC. It was just like, you know, they were asking me about Gemini. I was like, first of all, Bard's not good. And they're like, you mean, I'm like, yeah, the capabilities of consumer product is not up to date. So maybe the commoditization isn't going to be as fast. Okay, let's listen now to how confident Nadella is on OpenAI having sorted out its governance issues.

8:08So I'm comfortable. I have no issues with any structure. What we just want is good stability. And as I said, we don't even need. Like I'm not even interested in a board seat or we don't need control. We definitely don't have control. We have no, we just want to have a good commercial partnership and we want to be investors in the entity in even the way they're structured. So what I would like is good governance and real stability. That's it. You have a good governance and stability. What I would like. Isn't that what we all, what I would like. That's not what I have. Do you know one thing about those moments I always wonder?

8:44he had to anticipate the question. He might have even been given this topics beforehand. I feel there's got to be a better answer you come up with. Again, at this point to me, I don't know, does Microsoft, you continue deflecting these kinds of questions and say generic answers or do you commit to here's what this means for us and what we want from them? Well, I guess like the thing that I'm mostly surprised about is why he didn't say I'm confident the governance issues are in good hands right now obviously he doesn't think that so i i guess that like oh you know what i'm saying yep good nuance he did not yeah you're right what i would like never said they've moved past and we're looking forward to the future or something generically corporate like that he said what i would like so speaking of like some of the issues that we're going to see this year it will be this continued uh governance question with open ai and it kind of goes back to the question about Microsoft.

9:42It's like, what is their AI? Can their AI play exist without OpenAI? Okay, again, and this is going to be my umpteenth Google or Microsoft co-pilot reference already in this episode, but I think OpenAI is in trouble. I think OpenAI is in trouble. I think, again, ChatGPT Plus and that interface, it's a great product. It's been the leader in this. But when you start looking at Microsoft essentially distancing themselves from OpenAI, the same way AI has added theoretically a trillion dollars to Microsoft's market cap, you have to assume the Microsoft partnership has been instrumental in pushing OpenAI to the$80 to$100 billion valuation as well.

10:32You have the lawsuits from the New York Times. So from a copyright perspective, OpenAI is the poster child in Target. From a product standpoint, their own investor is giving away their product for free, which still blows my mind. I mean, at every level. And then from a governance and just corporate and organizational standpoint, they clearly have not really made people feel good about where they are. What's the bull case for OpenAI for you? Well, I think we're about to talk about it. But I'll say, first of all, I think Satya, I think he is in the fool me once, shame on you, fool me twice, shame on me mode, where he's like, all right, the first time we didn't see this coming, we will be prepared in case the second time comes and we're not going to bet the company on one anymore.

11:22And I think that's the gist of his statements here, because he was also like speaking to Brad and Brad asked him. This was this answer that he gave about governance was in Brad Stone from Bloomberg. The response that he gave was in response to a question about like, is he satisfied with the open AI government governance? And he tells Brad, like, you're speaking to Sam later, aren't you? Like, he has to figure that out before he gave this answer. So I found that really revealing. Now, okay, your question, what is the bull case for open AI? Let's just listen to Sam Altman. We're right about what's going to happen here.

11:57This is like bigger than just a technological revolution in some sense. I mean, sort of like all technological revolutions or societal revolutions, but this one feels like it can be much more of that than usual. And so it is going to become a social issue, a political issue. It already has in some ways, But I think it is strange to both of us that it's not more of that already. But with what we expect to happen this year, not with the election, but just with the increase in the capabilities of the products. And as people really catch up with what's going to happen, what is happening, what's already happened, there's like a lot of inertia always in society.

12:37So that is fascinating to me that Sam is basically telegraphing the fact that there are going to be enormous breakthroughs from OpenAI this year. And, you know, maybe some of it is CEO talk, but this has sort of become a theme for him. Talking about all the stuff that we're going to see this year is going to be like, you know, pretty groundbreaking stuff. And, you know, it doesn't seem like it's going to be like just a little bit smarter GPT model. I mean, what he's teasing is like something like he talks about that we're going to have to reckon with on a societal level. So that's the bull case for open AI, I think, is that they actually build that.

13:16I don't know. I think to me, the more I've looked, because again, I do believe and agree that the advancements and the speed of advancement is continuing again, even like mid journey V6 versus V5 is, you know, like a step change in how good the quality is and how easy it is to use. So the idea that maybe OpenAI is going to have some massive new release, I think, is very reasonable. Is it something that other people are not going to do? I don't know. Like, are other people catching up? The clods of the world, even MidJourney, other competitors of different parts of OpenAI's ecosystem. But again, anytime the conversation moves into these kind of like theoretical AGI Terminator-ish scenarios, you know, I start to question how much of that is just kind of like, as you said, CEO talk and promises versus reality.

14:16Again, we should definitely talk about the societal impact side of things. I think there's a lot of issues that are already being created by GPT-3 and then four, so eventually five. But I don't know. I wasn't sold on that. What does your gut tell you? Big stuff coming for OpenAI or that was CEO speak this year? Okay. Again, it's a tough one. And I think big stuff is coming, but I don't think it will be significantly different or better than what other companies are doing. Okay, I'll take the other side of that coin. Given what I'm hearing from Sam, I am very curious to see what technology they release.

14:59And there were some attempts at Davos to get him to talk about what's coming, but he very gently refused. Okay, but one more clip that we'll listen to is about the societal impact. And obviously the question is like whether this is going to take jobs. And I just spoke with Albert Wenger about that a couple of weeks ago, but, um, or last week, but this is something that was put to Sam, uh, front and center. And here's his response on how it's impacting coders. You know, you hear a coder say, okay, I'm like two times more productive, three times more productive, whatever than they used to be. And I like can never code again without this tool.

15:35You mostly hear that from the younger ones, but, um, it turns out, and I think this will be true for a lot of industries. The world just needs a lot more code than we have people to write right now. And so it's not like we run out of demand. It's that people can just do more. Expectations go up, but ability goes up too. Okay. I'm personally with Sam on this one as well. I do think that we are just going to see more capabilities and people who are using AI living up to the expectations that their companies have in terms of being able to ship. And I don't really see like, I don't see it as something that's going to really reduce wages or minimize jobs across the economy.

16:14What's your perspective? Oh, see, I think there will be a tremendous impact on jobs. I think it's going to reshape and reformulate a lot of different industries. I'm actually, I am still, I don't even know if the word is bullish, but convinced that that will happen. And I think, you know, there's going to be impact and there's going to be people needing to learn different skills and there's going to be, you know, it's going to be a big issue. But I did like that idea that the world will need more lines of code. That's pretty good. I think like from a pure like coding perspective, maybe it doesn't shrink things too much that people who understand, there's still going to always need to be people understand code and how to apply it.

17:01So and how to use AI systems to generate code. So I think that's reasonable. I mean, as a writer, I think I actually probably maybe the world does not need as many words. Oh, my God. It certainly does not. Exactly. Exactly. It needs better words, but not more words. Yeah. Words, the volume could decrease a little bit. So I think on that side, again, if you are a writer for an SEO optimized recipe site, that's probably going to change. change um i mean all those things that these things will all change i think but but i think from a societal impact i mean we have to talk about the upcoming u.s election we just the iowa caucuses were this monday like it's front and center right now and it's happening right now so i think that we yeah let's definitely get into this especially as it pertains to ai and obviously this is going to be a theme for us this year um both like the elections that are going to 70 to 80 democratic elections across the globe this week, sorry, this year, and of course, one in the US.

18:08We have a global audience. We won't go too crazy on US politics. But it is obviously a testing ground for a lot. This US presidential election is a testing ground for a lot of worries of technologies, new technologies. And of course, AI is going to be front and center. And I think there is this. So first of all, I'm glad people are concerned about the fact that AI could pose some problems in the election. But I think that the level of attention that's getting compared to the problem is maybe disproportionate. Maybe some, maybe people are trying to make up for the lack of attention for social media and then how that effort played and trying to say like AI will have an equal impact.

18:51And I just don't really see it. So let's just read the headline. So this is from the Wall Street Journal. OpenAI bans the use of AI tools for campaigning and voter suppression. And basically it says OpenAI's ChatGPT and DALI are some of the most powerful AI chatbot and image generation applications available. And then OpenAI said people aren't allowed to use its tools for political campaigning and lobbying. people also aren't allowed to create chatbots that impersonate candidates and other real people or chatbots that pretend to be local governments. I think that like, this is a thing. In campaigning, the cost to create media is so low that you don't really need AI to flood the world with disinformation and misinformation.

19:41Like it's going to be a problem with AI or without AI. and I don't think this dramatically causes a bigger problem. And I think that people, you know, if people are chatting with a politician chatbot, like, come on, like, is that going to sway an election? I doubt it. What do you think, Rajan? I don't know. So on one side, I do think the algorithms and you spoke with Kyle Chayka, Chayka, Chayka? Chayka. Chayka on Wednesday, I mean, it was a really good episode on his new book and the impact of algorithms. To me, that's a much greater issue than generative AI specifically, because, again, what is the distribution side of it is just as important, if not more important, than the actual content creation side.

20:28As you said, you know, you don't need generative AI to make up something and just tweet or post about it and then send it around. But I do think there's the disinformation side, then there's the kind of like complete fabrication and creation of a – the deepfake side. The idea that there was actually a politician back in December who had created – using generative AI. And it felt a little kind of PR gimmick-ish. It was Democrat Shemaine Daniels was running for Congress and created her own AI-generated robocalls. And the idea was she could make it more personalized and you could make it, you know, target people better and not have to do all the work of recording your own voice to target all these people.

21:16That to me was actually like an interesting, reasonable, productive use case of this. It's like actual tailored messaging to different constituencies that you use AI to generate so you don't have to sit in a studio all day to record wherever you would one records robocalls. But of course, you know, we are in for an endless world of deep fakes of Biden saying something or Trump saying something or whatever other candidate saying something and people believing it. And I think that is a capability that exists this year that has not existed in any previous election. yeah i'll tell you where i think this actually starts to resonate it's when video generation gets good and then when we can't tell the difference between um you know what a real video of something like donald trump saying is in a fake video oh wait oh sorry go go go yeah okay so i i guess like what i'm going to say on this one is and maybe maybe i should like i think what you're about to say is that already exists i would i'll say as a counter is when it it's easy for everybody to make it.

22:25And then when that happens, it's not going to be as much as people believing that these videos are the truth. It's going to be about people believing what they want to believe and not caring whether they're fake or real. That's going to start to cause some problems. All right. What do you have to say about that? Okay. So it already exists. You said I was going to say it and I said it. So there's a company, there's a product, Hey Jen, H-E-Y-G-E-N. And I've used it before. They kind of went viral and it's actually mind blowing that you can upload, you know, like a 30 to one minute, 30 second, one minute clip of yourself.

23:02And it will, originally they went viral, but it would translate into another language in your exact voice and relatively match your lips up to the new words that are being spoken. And I tested it. I was like having it have me speak Chinese and Spanish and showing it to native speakers. And they're actually like, this is not bad. And it actually kind of looked like it was me speaking. And so this morning there was a really interesting on Twitter. I opened it my entire for you feed was Javier Millet the new Argentinian firebrand president him giving a speech in Spanish that was translated in hey Jen to English so it's his voice it's his words it's his speech but it's now reaching a complete English language audience in English and and everyone was sharing it and obviously it was more about the content itself and this was not a deep fake in the sense of it was making things up about what he was saying.

24:02But it's honestly, it's so good that being able to change what people are saying and make it look reasonable and make it their lips look like they're kind of saying what there's, you're getting them to say in their voice. That's here. That's here this year. Definitely. Yeah. I saw you tweet about that and then I went and watched the video. It is really incredible how they can do those translations and make a video look like somebody is literally speaking the language in their voice, which is wild. Yeah, and honestly, the platform, you just hit translate. You specify the target language. It takes maybe for every minute of video, like 10 minutes.

24:42You can tell there's probably some serious processing going on on the other side, and then it looks good. It looks very good. So that technology is here, and we haven't even seen the first major mishap, And I think we will. Yeah. So maybe that's something that we see this year. The Davos consensus is what everybody at Davos starts to agree on, like all the elites, the prime ministers and the governmental officials and the business elites. They tend to walk away from the conference agreeing on something. And then it almost universally ends up being wrong. Like they showed up in January 2020, didn't speak at all about COVID.

25:22And of course, that's what dominated the world for the next couple of years. So Ben Smith was on Squawk Box. Ben Smith, the editor-in-chief of Semaphore, was on Squawk Box and talking about what the consensus is this year. And his perspective is that everybody at Davos this year is sure that Donald Trump is going to win the presidential election in the U.S. in 2024, which indicates, who knows, maybe that's good news for Biden or some other candidate that we don't even know about. Davos curse is the new Sports Illustrated curse for older listeners. So Apple has the Vision Pro coming out soon. It put the device in the hands of lots of influencers, reviewers.

26:05People took pains to say this is not a review. I just got a 30-minute demo from them. But we're already starting to see some reactions here about the direction of this device. And Ranjan, they are not good. So Bloomberg had a headline, big headline that said Apple's Vision Pro headset lacks blockbuster buzz and heated to energize shares. And basically saying that it's unlikely to play a major role in changing the perception that its growth is in the past. And this is from this is a quote from Denny Fish, who manages tech funds at Yanis Henderson. Anderson he says uh it's hard to ask someone to pay$3 ,500 for a product where there's a limit amount limited amount of content people can't get on their phone which means it'll be pretty darn niche at least for a couple years I think that's a great point have you seen the reviews of the Vision Pro I mean my perspective is you know I don't think it will necessarily be dead on arrival but it's really going to have a rough go of it as it gets released by Apple yeah I think what is a rough go is the big question because obviously you hear some people you know just make simple calculations at thirty five hundred dollars times x units suddenly you have a billion dollar new business but the reviews have been rough it's everyone's saying it's heavy everyone's saying that uh you know the typing does not work very well um it's like pinching no pinching is to select but they're supposed to be these virtual keyboards that you just like use your move your fingers and it types on the keyboard.

27:43Yeah, let me read German's quote from there. He says, Apple's, Mark German from Bloomberg, the Vision Pro virtual keyboard is a complete write-off, at least in 1.0. You have to poke each key one finger at a time like you did before you learned how to type. There is no magical in-air typing. You can also look at a character and pinch, you'll want a Bluetooth keyboard. Yeah, no, this is big. I mean, I think these issues are big. However, I think the one thing I read that seemed to be a good synopsis of what Apple's launch strategy is, because even though they've been giving out units and there's certainly been coverage and it's clear they've had fairly controlled demos with influential members of the press.

28:30And friendly members of the press. And friendly members of the press. that this is not a big launch. As crazy as that sounds to say, this is not some gigantic launch, that this is something where they're just trying to get this in the hands of the right people. In the end, it's going to be the developer ecosystem that creates the use cases. It's going to be the companies that build the right apps and experiences on it. Some number of people will try it out and talk about how cool it is to their friends who eventually will get a version two. I think that not overblowing the potential of this at the start is actually the right move.

29:12Okay, so I actually completely agree with you on that point, believe it or not. I just wonder whether there is a way for them to not blow out the importance. Like, it is going to have to be a very, very managed takeoff if that's the case. Because think about the release event that they did about this, right? not just a device, a new form of computing. So how do you go from that showing all these like magical scenarios now to be like, all right, this is a half launch, patience, please. Yeah, no, that's fair. Also, let's not forget, remember the launch event, the dad who was wearing it and like staring at his two kids, like - It was very sad, yes.