Will AI Save The World + Azure's Origins — With Bob Muglia

0:00LinkedIn presents.

0:13Welcome to Big Technology Podcast. a show for cool-headed, nuanced conversation of the tech world and beyond. Bob Muglia is our guest today. He's the former CEO of Snowflake, a$50 billion plus technology company. And he was also the head of server and tools at Microsoft, which is the division that produced Azure, Microsoft's cloud services offering. And it's also the division that its current CEO, Cy Inadella, came out of. Muglia has a new book out called The Datapreneurs, The Promise of AI and the Creators Building Our Future. and he's here to talk about it. But first, we're going to dive into Mark Andreessen's new essay, Why AI Will Save the World, and dive into its arguments and merits.

0:53In the second half, Muglia and I will have a discussion I've been looking forward to having for a long time. We're going to speak about the formation of Azure and the conflicts inside Microsoft before it emerged. This is definitely one of the tech stories I've wanted to get to the bottom of, and we're going to hear it from Muglia, who is right there. So here we go. My conversation with Bob Muglia is coming up right after this. Capital One's tech team isn't just talking about multi-agentic AI. They already deployed one. It's called Chat Concierge, and it's simplifying car shopping. Using self-reflection and layered reasoning with live API checks, it doesn't just help buyers find a car they love.

1:32It helps schedule a test drive, get pre-approved for financing, and estimate trade in value. Advanced, intuitive, and deployed. That's how they stack. That's technology at Capital One. The truth is AI security is identity security. An AI agent isn't just a piece of code. It's a first-class citizen in your digital ecosystem, and it needs to be treated like one. That's why Okta is taking the lead to secure these AI agents. The key to unlocking this new layer of protection? An identity security fabric. Organizations need a unified, comprehensive approach that protects every identity, human or machine, with consistent policies and oversight.

2:10Don't wait for a security incident to realize your AI agents are a massive blind spot. Learn how Okta's identity security fabric can help you secure the next generation of identities, including your AI agents. Visit Okta.com. That's O-K-T-A dot com. Bob, welcome to the show. Good to be here, Ox. Great to have you. So I emailed you in March 2019, and I was writing this book about Microsoft, actually all the tech giants, but specifically a focus on reinvention at Microsoft. You ran the server and tools business there, Satya Nadella, eventually took it over and transitioned it to a cloud business.

2:49And we're going to talk about that. We'll start with AI. We'll talk about that. And I never heard back, and I almost feel bad for sending it at that time because I I think you were in the middle of being pushed out as leader of Snowflake. I was transitioning out of Snowflake. And I don't remember the email. I apologize for that. I try and respond to these things when I don't remember, but I apologize for that if I didn't get back to that. No apology necessary. And when I saw your name come through that you had a book in the works, I was like, all right, this is finally the conversation with Bob.

3:19Here we are. I'm glad we have a chance to do it now. Of course. And you're at a very interesting point. You're releasing a book. You actually wrote the book, had some parts about ChatGPT and Dolly in it with basically a month after they came out. And now it's being published this month. And it's been amazing what's happened with your previous company, Microsoft, Snowflake's obviously involved. And the discussion has just evolved in so many different ways. So in your book, towards the end, you have some discussions about AI ethics and sort of the rules we want to set for AI. And recently, Mark Andreessen, the head of the VC firm Andreessen Horowitz, who I'm sure our listeners are familiar with, came out with this post and basically said, we don't really need to worry about any of the worst case scenarios about AI.

4:10I mean, I'm simplifying it a bit, but point by point, taking the AI doomerist's concerns and basically saying these are actually fallacies. I know you read the post, so let's just start there. I'm curious what you thought about Andreessen's central thesis. Well, generally speaking, I think I agree with Mark. Let me start by saying that. I mean, in the sense that I am an AI optimist, and he is also an AI optimist. I'm a technology optimist, and clearly Mark is as well. And I think that AI is going to do incredibly positive things for the world, and we're going to see an incredible set of advances over the next few years with these co-pilots or assistants that are going to help us in so many ways.

4:50AI is going to make us smarter and more productive, and that's a good thing. And in general, I agree with Mark that there's a lot of reason to be optimistic about jobs as well. Although Mark did over, you know, he did simplify it a little bit in that, in that while technology has historically always created more jobs than it's displaced, it's very inconvenient and challenging, you know, really the problem actually for the people that are displaced. And I don't want to, and to me, those are the sorts of issues that are still going to be very real societal issues over time. Even if Mark is fully correct in that more jobs are created, we'll have to deal with the reality that many people will lose some jobs.

5:29And I think we've seen that over time. Perhaps it's a great – the transition goes well and there's great jobs ahead for everyone, but I'm hopeful about that. Wait, wait. So let me stop you there for a minute because we are definitely going to get to the jobs situation. But, and I'm going to start to read some of these arguments that Andreessen makes and actually want to get your perspective on it. Yeah, sure. Because he made a lot of different arguments. Oh, yes, he did. He was very, he was very thoughtful in what he put together. Exactly. So he actually splits the categories of the people that are criticizing AI into two different categories.

6:00And I think it's very interesting. He's shaking your head. I'm going to read it anyway. He calls it the Baptists and the bootleggers. And the Baptists, he say, are the people that actually have serious concern and really believe in it. And the bootleggers, he says, are people who are kind of doing this to profit off of the fear, to basically say that AI is dangerous, to say that AI is going to take jobs and to put themselves cynically in a better position. So, and he says something very interesting. He says the bootleggers always win because the Baptists are just going to raise the alarm and be of, you know, be concerned while the bootleggers with their malicious intent are going to go and take that fear and actually turn it into something that furthers their self-interest.

6:46What do you think about that? Well, I mean, Mark sort of talks about two extremes of people's approaches to things. And he's not wrong about what he's saying, what I would say. But I think he oversimplified that a wee bit. It's a little bit too simplistic. I have a slightly different, what I would say is, and I think Mark referenced this, but not explicitly. And I think it's really worth referencing this explicitly and distinguishing between two kinds of fears that people have about AI. One is the fear that AI, which is a tool that people will use, and people will use it for bad things. And Mark very clearly said in his piece that that's going to happen, and I agree with him on that.

7:29In general, we have built a lot. As humanity, we built tools from the very beginning. Tools have enabled us to be as productive as we are in society, and obviously, it's how we live our lives. And many tools have positive and negative uses, and AI is absolutely one of those tools. that can be used for ill and can be used for a lot of good. And AI is a very powerful tool. And so the good and the evil could be very powerful. But I have a lot of belief that we will fully control that, that the right things will happen. Marx says that no new legislation is required. I'm not quite as optimistic about that as he is, that no new legislation is required.

8:06I think there may need to be some. But I do share some of Marx's concerns that we could over-legislate this and put too many restrictions in. And the worst concern I would have, and Mark reflects this in his blog, is that we restrict all of the innovation that is happening in the open source community right now. Because I think that's where the most interesting innovation is happening in AI. And I think it would be a real bad mistake if through perhaps well-intentioned, but misapplied government regulation makes it so that only the largest companies have the money to deal with all of the regulatory concerns.

8:48So I do have some of his concerns on that, but this is AI being used by people as a tool. Exactly. And the second thing that Mark addressed, and he kind of waved this off, and I'm not with him on this, is this idea that AI will eventually become an intelligent entity that is effectively the equal of us. Mark says that's not going to happen. He certainly seems to imply that that's not going to happen. I'm not so sure about that. And when I talk to a bunch of the AI researchers, they're not so sure about that either. So that's an interesting question. And that's where I think the bigger and broader concerns of existential risk come in.

9:29This is not a concern, I think, in the next five or even 10 years. But if you go out longer, as AI gets smarter and smarter. Could it do some things to deceive us? Are there ways we could lose control of it? While I am an optimist that we will not lose control and that we can always pull the plug because these things are, in fact, digital circuits running in data centers that at least for now we still control. I do think there is something that needs to be thought about there. So that's one area where I disagree with Mark. Yeah. And I think he does say that they are going to be, I don't think he precludes the idea that they're going to be on par with us intellectually.

10:06I just think that he believes that there is, you know, a kill switch because you can unplug it the same way you can unplug a toaster. And let me just read his argument to what you would say and sort of we can go back because, by the way, like, I think what he's trying to say broadly by writing this piece is that startups, you know, because that's who he funds, are going to be hurt by this fear, which he calls a moral panic, because what's going to happen is the established companies will use it to legislate barriers and then eventually lead that will lead to regulatory capture. I'm sure you, you know, who ran a company, basically from close to no revenue to something that's now, you know, just a few years later, $59 billion market cap.

10:48You understand the idea that the bigger companies want to box out the smaller companies by manipulating these fears. Absolutely. And I also understand as a very large company, what can happen when government comes down on them? Because I was very much at Microsoft and part of Microsoft during the DOJ lawsuit. So I'm quite familiar with the regulatory regime and some of the downsides associated with it. And so I have, again, I share some of those concerns with Mark, and I think it would be terrible if that happened. And I don't think there's a reason for it to happen either. But it can happen. I was just talking with someone today, just like these things logically might not make sense, right?

11:25Regulation that boxes out smaller companies might not make sense, but narratives are really important here because they are the stories that end up, you know, either pushing something forward or getting it blocked, and then you can end up with suboptimal outcomes, even though there's like consensus. Consensus means nothing when it comes to government and regulation. I do agree. There's concern about this. I think that when we really want to focus on issues with AI, I am much more concerned not about what's going to happen in the next five years, but will happen, say, 10 to 15 or 20 years from now as these things get smarter and smarter.

12:07And we have to think about how to treat them and how we work with them. Because I do think they'll get smarter. And while I do agree that we can pull the plug, eventually maybe we can't. I mean, there may come a day when they control enough of what they're doing that we won't be able to pull the plug. And that's why I do think we need to make sure that their values are aligned with what society believes in general are appropriate values. Values are complicated. Everybody has different values. But there are societal values and societal norms. And it's appropriate as we create these things, and we are creating them, that as they get smarter and smarter, we make sure that they accept the norms that humans think are appropriate.

12:48Which is why I think that Asimov's laws are quite relevant in the conversation about AI that is AGI and potentially super intelligent AI. Exactly. And he does go through some of the arguments about why the fact, like, let me just read his argument to you in terms of why he believes the fact that we might lose control of this to be not feasible. So he basically says AI is math. And he goes, my response to their position is non-scientific. What is that their position is non-scientific? What is the testable hypothesis? What would falsify the hypothesis? How do we know when we are getting into a danger zone?

13:30These questions go mainly unanswered apart from you can't prove it won't happen. And in fact, he says their position is so non-scientific and so extreme, a conspiracy theory about math and code, and some are already calling for physical violence. that I will do something I normally do not do and question their motives as well. Just respond a little bit to what he's saying in that part. Well, first of all, I am heartened that there is an outcry of people that are asking questions about this. I think that's a good thing. And the conversation that we've been having across society for the last five months, six months, because it really all started in December, right after ChatGPT shipped, that's when this conversation got into high gear.

14:15I think it's a helpful conversation. The concerns about regulation that Mark has, I think, are all very relevant. And while I do recommend that we be thoughtful about this, and in particular as the very large language models that get developed, say GPT-5 as an example, and it's ilk that are coming out from the other tech companies, that we think through what those capabilities are and we be thoughtful about how they benefit all of us and will not take us down these negative directions. I do think all of those conversations are important. I also worry about the potential for over-regulation and over-reaction.

14:52And it's sort of funny because my position on this is shifting relatively rapidly as time goes on, because in January, I was worried that people wouldn't think this was important enough. And I'm clearly no longer have that concern because there's been so many articles and so much outcry about this. And frankly, so many scientists that really deep into the technology, expressing their concerns. Now, Mark can assign motives for this or not. I'm not trying to assign any motives associated with it. I think people are just genuinely concerned about what they're doing with the realization that there are some incredible, that this is very, very powerful technology that will affect society.

15:32That's why I always come back to saying, you know, there's a lot to learn from Isaac Asimov and the things that he talked about, because he really did project a society where we coexisted with intelligent machines. And he thought about it in his stories. He wrote about many of the challenges that are associated with that. They're really parables that provide us with some guidance here. And so I think that's all very relevant. I do think that it's appropriate for government to be worried about this and talking about this. Should regulation be going down now? I think it's a little early. I definitely think it's early because I don't think we know what to regulate.

16:07And I do think we have to be cautious. I think we'll learn a lot as well from what other parts of the world are doing. Europe has historically taken a more active regulatory role in technology. And it would not surprise me if that's also the case here. And we already saw Italy make some moves in that regard. Well, they rolled that back though. So let's talk about, so you mentioned the job thing. So let's talk about the job thing for a second. Cause I like, I do want to, let's get into this discussion a little bit about his arguments against being worried. I'm not, I'm not saying I'm taking a side.

16:36I just think it's interesting to discuss. So he talks about how automation, the automation can kill jobs. And he goes, the core mistake of the automation kills jobs, doomers, that they doomers keep making is called the lump of labor fallacy. This fallacy is incorrect notion that there is a fixed amount of labor to be done in the economy at any given time. Either machines do it or people do it. And if machines do it, there will be no work for people to do. And he goes, when technology is applied to production, you get productivity growth, an increase in output generated by a reduction in inputs.

17:09The result is lower prices for goods and services. As prices for goods and services fall, we pay less for them, meaning that we now have extra spending power, which we would use to buy things. This increases demand in the economy which drives the creation of new production including new products and new industries which then creates new jobs for the people who were replaced by machines and prior jobs. The result is a larger economy with a higher material prosperity, more industries, more products and more jobs. Your thoughts? I agree with all that. I mean, he's totally right. The only thing is this one element he left out which is time.

17:45Yeah. Yeah. And these things happen over a period of time. And what's happening now, and this is an incredibly important point, is they're happening very quickly. In the past, when technology affected jobs, it happened over years and decades. It's not going to take decades. It's just not. It's months, it's days, months, and maybe years we're talking about. And so if in fact we have automation to the point where autonomous self-driving cars become a reality. And Uber and all of the other car fleets replace their drivers with robots. That is going to displace the jobs of a lot of people that use that particular field as almost either it's their primary income or it's a last ditch income.

18:37So there's real human displacement issues that I think are quite real and cannot be ignored in this. I mean, it's very, very important. Over time, Mark is correct. I believe more jobs will be created, but will those people be able to be trained to do some of those jobs? Classically, there's always been an issue in human history with this and the speed at which the potential for this displacement is happening makes this more acute than perhaps the past ones in the past. That I agree with 100%. The way that he ends his piece, I mean, he ends his piece with this like cute thing about how the people who build the AI before legends and the people who are building AI today are heroes, which is like kind of an interesting assignment to people without, you know, to say that each and every one of them fit into that category is, I think it cheapens the word legend and hero, but that's a lecture for another day.

19:29I'm curious. That's just the ending of his piece though. I mean, you know, it's a little cute ending. I'm curious, he makes a point right before that talking about how the US and China are basically interlinked in a war over AI. And his theory is basically go ahead and win from the US standpoint. Your perspective on that? I think that then, you know, well, first of all, I'm worried, you know, again, I said, I have no, I have some concerns about of an existential nature, you know, in a long-term period as AI gains super intelligence in particular, assuming that that happens. No guarantee it happens, but it sure looks like it's going to happen.

20:09I have bigger concerns, honestly, about the situation with China and the United States. I think, you know, I'm a Cold War baby, right? I was born in the 1950s. I ducked and covered when I was a child, right, in elementary school. I mean, I remember the nuclear fear that we had. I grew up with that over my life, you know, is one of the things that people were afraid of. It's been almost a non-issue since 1990, after the Berlin Wall fell. We've really forgotten about what that means to live in a world where we have these large existential enemies, because that is an existential situation. Nuclear war is an existential thing.

20:56The Taiwan situation is an extremely scary situation right now, and it has a potential to put us in conflict with China. Right, but the broader AI thing is sort of... The AI fits into all of this, though, because in a way, AI is the solution through all of this, is that if it progresses humanity forward collectively. I generally speaking agree, however, with Mark, that we must progress AI quickly because the Chinese, they will have a different approach to this. As I say, values are critical. values will determine what is created. We are creating these systems, whether they're tools or entities, we are creating them and they have the values that we put in them.

21:43And there's no question that China will have a different set of values. So I do agree that we need to move quickly. I hope that through all of this, AI will help to bring people together, humanity across the world together, because this conflict that I see coming with China doesn't seem to help anybody. And it does make me very concerned. Okay, I'm with you on that also. Throughout this conversation, you've said that you think AAGI is artificial general intelligence is inevitable. And it certainly seems that way in your book. So I'm curious what makes you so sure of that. And, you know, is that something we actually want?

22:21Well, first of all, whether we want it or not may not matter because I think we're on a trajectory to get it. No, no, of course it matters. It definitely matters. You can't just throw up your hands and say it doesn't matter. We built them, though. Yes, we did. Yes, we did. And just like we will build AGIs if we have the ability to do so. I think that we need it. When we're working towards technological progress, you can't just throw up your hands and say it's going to happen. This is a decision that people make. People is a very broad statement, right? I mean, you can have your decision. Look, all these scientists signed that thing that said we should put a six-month moratorium on AI.

23:04It went nowhere, absolutely nowhere. People ignored it completely. I mean, short of government regulation, which I completely do not believe is appropriate. I can believe that would be disastrous. I don't know how to stop it. I don't know how to stop it. Now, you can tell me, you can disagree. And smart people can say, hey, this trajectory doesn't result in what one would call an AGI, which is a system that has a computer system that has the intelligence of a median human. That's kind of the general definition of an AGI. And I think we're proceeding on a very well-defined path towards that right now.

23:42It's certainly not provable that it's going to happen. And I don't have proof that it'll happen. And I will also tell you that when I started writing the book, I had no idea that it would happen in my lifetime. Okay. But we still don't know it will happen. But now I believe it. The difference is I lived over 60 years of my life having spent a lot of time reading science fiction as a kid in Asimov, spending a lot of time thinking about intelligent robots. And my entire career has been about making computer systems more intelligent and more productive for people. and I've been able to use the tools that are at my disposal, the chips and the power that they give, the database software, the application software, the development tools, the languages, all that stuff to be able to build, to work with teams to build some really cool products over time.

24:32Now, intelligence is available. For God's sakes, computers can respond effectively to English. apis english is now an api oh my gosh my head explodes it's stuff like this so so what happened to be straight and it and it's it's you know it was i'll say you can you can disagree with it if you you know if you like but it's what i believe i i spent 60 plus years of my life believing that all this technology was interesting but i wouldn't see a world where intelligence was actually available to be applied to a broad set of problems. And now it is. Will that result in an AGI? Given the trajectory we're on, I totally believe it.

25:11Will it result in superintelligence? That's a fair question. That's a fair question. And then does it result in what people sometimes refer to as a technological singularity, which is an acceleration of progress kind of beyond anything we can imagine? And no one knows that, of course. No one knows that. Was it just ChatGPT and Dolly that convinced you? It was all the pieces. The first thing that got me honestly was, was, was the, was the, was co-pilot, Microsoft co-pilot. Okay. Cause that codes alongside you. And, and when I realized that, that of all the things that we expected, you know, the, the, the technology advances to, to drive productivity of developers like that was just, it, people that came out of left field from my perspective, that all of a sudden a computer could assist in writing code such to the point that a developer could get 40 % or more of their work done by a co-pilot AI machine.

26:09I didn't see that coming. I do speak with CEOs. I guess you've seen it in action. I've spoken with some CEOs who said, yeah, we all use it. And I say, well, what sort of productivity increase have you seen? And they're like, still waiting. But the CEOs that you speak with, are they in that same place or when it comes to co-pilot in particular are they in that same boat or have they actually seen that 40 percent increase what i hear from the microsoft people is is that people are using it all over the place and that it is you know that a lot of the code that's being written is written by by this um by the ai by by co-pilot now it's certainly true that most of what co-pilot is writing today is the lower value code right right that that doesn't require the level of intelligence but you know but but that's still an incredible productivity improver and you know we haven't even begun to see these you know right now what we have are no you're setting you're setting me up for for what i'm going to come to which is the will i ai save the world question that that andreese i mean andreese's post is will ai why ai will save the world so i'm going to give you a minute to speak about that in the moment but the ai uh it's very interesting i've never heard someone speak uh refer to ai as um sorry artificial general intelligence as a median human intelligence.