YouSaid · the spoken record

Mark Zuckerberg

lines on the record
172
first
2023-06-08
most recent
2023-06-08
sittings or episodes
1
sources
podcast

Every line below is reproduced as it was said and linked to the record it came from. Nothing here is summarised or generated. Directory · Search · Corrections

  1. “Kind of have the trust and have a sense of how the company works, can they go work remotely now if they want and still do it as effectively? And we've done all these studies that show it's like, okay, does that affect their performance? It does not. But for the new folks who are joining and for people who are earlier in their career and need to learn how to solve certain problems and need to get ramped up on the culture, when you're working through really complicated problems where you don't just want to sit in the you don't just want the formal meeting, but you want to be able to brainstorm when you're walking in the hallway together after the meeting, I don't know. It's like we just haven't replaced the kind of in-person. Dynamics there yet with anything remote yet.”

    2023-06-08 · Lex Fridman Podcast · #383 – Mark Zuckerberg: Future of AI at Meta, Facebook, Instagram, and WhatsApp · IDENTIFIED FROM THE TRANSCRIPT · source

  2. “Yeah, exactly. It's like all the metaverse work and the ability to be. Feel like you're truly present no matter where you are. I think once you have that all dialed in, then we may one day reach a point where it really just doesn't matter as much where you are physically. I don't know. Today it still does, right? So yeah, for people who, there are all these people who have special skills and want to live in a place where we don't have an office, are we better off having them at the company? Absolutely. And are a lot of people who work at the company for several years and then build up the relationships internally?”

    2023-06-08 · Lex Fridman Podcast · #383 – Mark Zuckerberg: Future of AI at Meta, Facebook, Instagram, and WhatsApp · IDENTIFIED FROM THE TRANSCRIPT · source

  3. “Yeah, I mean, I think it's a thing that's here to stay. But I think that there's value in both, right? It's not, um, I wouldn't want to run a fully remote company yet at least. I think there's an asterisk on that, which is that.”

    2023-06-08 · Lex Fridman Podcast · #383 – Mark Zuckerberg: Future of AI at Meta, Facebook, Instagram, and WhatsApp · IDENTIFIED FROM THE TRANSCRIPT · source

  4. “Do a great job here, you know that they're going to be great if they came and joined full time. And that's one of the reasons why we've invested so much internship is basically just it's a very useful sorting function, both for us and for the people who want to try out the company.”

    2023-06-08 · Lex Fridman Podcast · #383 – Mark Zuckerberg: Future of AI at Meta, Facebook, Instagram, and WhatsApp · IDENTIFIED FROM THE TRANSCRIPT · source

  5. “Okay, in an organization that's not growing as quickly, the questions might be a little different though. And there you asked about young people specifically, like people out of college. And one of the things that we see is it's a pretty basic lesson, but we have a much better sense of who the best people are who have interned at the company for a couple of months than by looking at kind of a resume or a short interview loop. Obviously, the in-person feel that you get from someone probably tells you more than the resume. And you can do some basic skills assessment. But a lot of the stuff really just is cultural. People thrive in different environments. And on different teams, even within a specific company, and it's like the people who come for even a short period of time over a summer.”

    2023-06-08 · Lex Fridman Podcast · #383 – Mark Zuckerberg: Future of AI at Meta, Facebook, Instagram, and WhatsApp · IDENTIFIED FROM THE TRANSCRIPT · source

  6. “I think that that kind of works, and that's basically how I've tried to build my team. I'm not in a rush to not be running the company, but I think in an alternate universe where one of these other folks was running the company, I'd be happy to work for them. I feel like I'd learn from them. I respect their kind of general judgment. They're all very insightful. They have good values. And I think that that gives you some rubric for You can apply that at every layer. And I think if you apply that at every layer in the organization, then you'll have a pretty strong organization.”

    2023-06-08 · Lex Fridman Podcast · #383 – Mark Zuckerberg: Future of AI at Meta, Facebook, Instagram, and WhatsApp · IDENTIFIED FROM THE TRANSCRIPT · source

  7. “So there are lots of different cuts on this question. I mean, I think when an organization is growing quickly, one of the big questions that teams face is do I hire this person who's in front of me now because they seem good? Or do I hold out to get someone who's even better? And the heuristic that I always focused on for myself and my own kind of direct hiring that I think works when you recurse it through the organization is that you should only hire someone to be on your team if you would be happy working for them in an alternate universe”

    2023-06-08 · Lex Fridman Podcast · #383 – Mark Zuckerberg: Future of AI at Meta, Facebook, Instagram, and WhatsApp · IDENTIFIED FROM THE TRANSCRIPT · source

  8. “Is it as valuable to have a lot of managers who have extra capacity waiting for new people? No, right? So now we can sort of defragment the organization and get to a place where the average is closer to that seven or eight. And it just ends up being somewhat more kind of compact management structure, which decreases the latency on information going up and down the chain. And I think empowers people more. But I mean, that's an example that I think it doesn't kind of undervalue the importance of management and the kind of the personal growth or coaching that people need in order to do their jobs well. It's just, I think, realistically, we're just not going to hire as many people going forward. So I think that you need a different structure.”

    2023-06-08 · Lex Fridman Podcast · #383 – Mark Zuckerberg: Future of AI at Meta, Facebook, Instagram, and WhatsApp · IDENTIFIED FROM THE TRANSCRIPT · source

  9. “Then that means that you need managers who have capacity to onboard new people. And also, if you have a new manager, you may not want to have them have seven direct reports immediately because you want them to ramp up. But the thing is going forward, I don't want us to actually hire that many people that quickly, right? So I actually think we'll just do better work if we have more constraints and we're leaner as an organization. In a world where we're not adding so many people as quickly.”

    2023-06-08 · Lex Fridman Podcast · #383 – Mark Zuckerberg: Future of AI at Meta, Facebook, Instagram, and WhatsApp · IDENTIFIED FROM THE TRANSCRIPT · source

  10. “Well, I mean, I believe a lot in management. I think there are some people who think that it doesn't matter as much, but look, I mean, we have a lot of younger people at the company for whom this is their first job. And, you know, people need to grow and learn in their career. And that all that stuff is important. But here's one mathematical way to look at it. At the beginning of this, we I asked our people team what was the average number of reports that a manager had? And I think it was around three, maybe three to four, but closer to three. I was like, wow, like a manager can best practices that person can manage seven or eight people. But there was a reason why it was closer to three. It was because we were growing so quickly. And when you're hiring so many people so quickly”

    2023-06-08 · Lex Fridman Podcast · #383 – Mark Zuckerberg: Future of AI at Meta, Facebook, Instagram, and WhatsApp · IDENTIFIED FROM THE TRANSCRIPT · source

  11. “And if you try to do good work but you don't have the right marketing team or the right legal advice, make some pretty big blunders. But at the same time, if you have too big of things and some of these support roles, then that might make it so things are just move a lot. Maybe you're too conservative or you move a lot slower than you should otherwise. Those are just examples. But”

    2023-06-08 · Lex Fridman Podcast · #383 – Mark Zuckerberg: Future of AI at Meta, Facebook, Instagram, and WhatsApp · IDENTIFIED FROM THE TRANSCRIPT · source

  12. “Latency of information that you get. We talked about this a bit earlier in terms of kind of the joy of and the feedback that you get doing something like Jujitsu compared to running a long-term project. But I actually think part of the art of running a company is trying to constantly re-engineer it so that your feedback loops get shorter so you can learn faster. And part of the way that you do that is by I kind of think that every layer that you have in the organization means that information might not need to get reviewed before it goes to you. And I think making sense that the people doing the work are as close as possible to you as possible is pretty important. So there's that. And I think over time, companies just build up very large support functions that are not doing the kind of core technical work. And those functions are very important. But I think having them in the right proportion is important.”

    2023-06-08 · Lex Fridman Podcast · #383 – Mark Zuckerberg: Future of AI at Meta, Facebook, Instagram, and WhatsApp · IDENTIFIED FROM THE TRANSCRIPT · source

  13. “Yeah, I mean, I wanted to. I want to empower Engineers more, the people are building things, the technical teams. Part of that is making sure that the people who are building things aren't just at like the leaf nodes of the organization. I don't want like eight levels of management and then the people actually doing the work. So we made changes to make it so that you have individual contributor engineers reporting at almost every level up the stack, which I think is important.”

    2023-06-08 · Lex Fridman Podcast · #383 – Mark Zuckerberg: Future of AI at Meta, Facebook, Instagram, and WhatsApp · IDENTIFIED FROM THE TRANSCRIPT · source

  14. “Pretty big thrash that we had seen over the last 18 months. Some of it macroeconomic induced, some of it specific, some of it competitively induced, some of it just because of bad decisions, right? Or things that we got wrong. I decided that we needed to get to a point where we were a lot leaner. But look, I mean, but then, okay, it's one thing to do that, to decide that at a high level, then the question is, how do you execute that as compassionately as possible? And there's no good way. There's no perfect way for sure. And it's going to be tough no matter what. But as a leadership team here, we've certainly spent a lot of time just thinking, okay, given that this is a thing that sucks, is the most compassionate way that we can do this? And that's what we've tried to do.”

    2023-06-08 · Lex Fridman Podcast · #383 – Mark Zuckerberg: Future of AI at Meta, Facebook, Instagram, and WhatsApp · IDENTIFIED FROM THE TRANSCRIPT · source

  15. “Reasons that aren't connected to their own performance or the culture not being a fit at that point, it's really just a kind of strategy decision and sometimes financially required, but not fully in our case, especially on the changes that we made this year. A lot of it was more kind of culturally and strategically driven by this push where I wanted us to become a stronger technology company with more of a focus on building more technical and more of a focus on building higher quality products faster. And I just view the external world as quite volatile right now. And I want to make sure that we had a stable position to be able to continue investing in these long-term ambitious projects that we have around continuing to push AI forward and continuing to push forward. All the metaverse work. And in order to do that, in light of the”

    2023-06-08 · Lex Fridman Podcast · #383 – Mark Zuckerberg: Future of AI at Meta, Facebook, Instagram, and WhatsApp · IDENTIFIED FROM THE TRANSCRIPT · source

  16. “Yeah, I mean, that's it That's it. I mean, you basically have a significant number of people who this is just not the end of their time at Meta that they or Or I would have hoped for when they joined the company. I mean, running a company, people are constantly joining and leaving the company for different directions, but for different reasons. Layoffs. Uniquely challenging and tough in that you have a lot of people leaving”

    2023-06-08 · Lex Fridman Podcast · #383 – Mark Zuckerberg: Future of AI at Meta, Facebook, Instagram, and WhatsApp · IDENTIFIED FROM THE TRANSCRIPT · source

  17. “The ability to think through how to shape their organizations in a way that hopefully can be good for the industry and make all these companies more productive over time. So I think that that was one where I think he was quite ahead of a bunch of the other companies on. And what he was doing there, again, from the outside, very hard to know. It's like, okay, did he cut too much? Did he knock enough? Whatever. I don't think it's like my place to opine on that. And you asked for a positive framing of the question of what do I admire? What do I think it went well? But I think that certainly his actions. Led me and I think a lot of other folks in the industry to think about hey, are we kind of doing this as much as we should? Could we make our companies better by pushing on some of these same principles?”

    2023-06-08 · Lex Fridman Podcast · #383 – Mark Zuckerberg: Future of AI at Meta, Facebook, Instagram, and WhatsApp · IDENTIFIED FROM THE TRANSCRIPT · source

  18. “I think that those were generally good changes. And I'm also, I also think that it was probably good for the industry that he made those changes, because my sense is that there were a lot of other people who thought that those were good changes, but who may have been a little shy about doing them. And I think he, you know, just in my conversations with other founders and how people have reacted to the things that we've done, what I've heard from a lot of folks is just, hey, when someone like you, when I wrote the letter outlining the organizational changes that I wanted to make back in March and when people see what Elon is doing, I think that that gives people.”

    2023-06-08 · Lex Fridman Podcast · #383 – Mark Zuckerberg: Future of AI at Meta, Facebook, Instagram, and WhatsApp · IDENTIFIED FROM THE TRANSCRIPT · source

  19. “It's like you can agree or disagree with exactly all the tactics and how we did that. Obviously, every leader has their own style for if they need to make dramatic changes for that, how you're going to execute it. A lot of the specific principles that he pushed on around basically trying to make the organization more technical, around decreasing the distance between engineers of the company and him like fewer layers of management.”

    2023-06-08 · Lex Fridman Podcast · #383 – Mark Zuckerberg: Future of AI at Meta, Facebook, Instagram, and WhatsApp · IDENTIFIED FROM THE TRANSCRIPT · source

  20. “Gosh, it's always very difficult to offer specific critiques from the outside before you get into this because I think one thing that I've learned is that everyone has opinions on what you should do and running the company you see a lot of specific nuances on things that are not apparent externally. And I often Think that some of the discourse around us would be could be better if there was more. Of space for acknowledging that there's certain things that we're seeing internally that guide what we're doing, but I don't know. I mean, since you asked what is going well. Know, I do think that Elon led a push early on to make Twitter a lot leaner. And I think that that”

    2023-06-08 · Lex Fridman Podcast · #383 – Mark Zuckerberg: Future of AI at Meta, Facebook, Instagram, and WhatsApp · IDENTIFIED FROM THE TRANSCRIPT · source

  21. “I've just been pretty fascinated by that and seeing how that works. To some degree, end-to-end encryption on WhatsApp and as we bring it to other services provides an element of it because it pushes the service really out to the edges. I mean, the server part of this that we run for WhatsApp is relatively very thin compared to what we do on Facebook or Instagram and much more of the complexity is and how the apps kind of negotiate with each other to pass information in a fully end-to-end encrypted way. But I don't know. I think that that is a good model. I think it puts more power in individuals' hands. And there are a lot of benefits of it if you can make it happen. Again, this is all pretty speculative. I mean, I think that it's hard from the outside to know why anything does or doesn't work until you kind of take a run at it. And so I.”

    2023-06-08 · Lex Fridman Podcast · #383 – Mark Zuckerberg: Future of AI at Meta, Facebook, Instagram, and WhatsApp · IDENTIFIED FROM THE TRANSCRIPT · source

  22. “Execute on, and others have too, right? I mean, byte dunes obviously have done a very good job with TikTok and have reached more than a billion people there. But it's certainly not automatic. I think you need a certain level of execution to basically get there. And I think for whatever reason, I think Twitter has this great idea and sort of magic in the service. kind of cracked that piece yet. And I think that that's made it so that you're seeing all these other things, whether it's Mastodon or Blue Sky that I think are maybe just different cuts at the same thing. But I think through the last generation of social media overall, one of the interesting experiments that I think should get run at larger scale is what happens if there's somewhat more decentralized control. And if it's like the stack is more open throughout.”

    2023-06-08 · Lex Fridman Podcast · #383 – Mark Zuckerberg: Future of AI at Meta, Facebook, Instagram, and WhatsApp · IDENTIFIED FROM THE TRANSCRIPT · source

  23. “Well, I just think it's hard to build these companies. So it's not that every idea automatically goes and gets a billion people. It's just that I think that that idea coupled with good execution should get there. But I mean, look, we hit certain thresholds over time where we kind of plateaued early on and it wasn't clear that we were ever going to reach 100 million people on Facebook. And then we got really good at dialing in internationalization and helping the service grow in different countries. And that was like a whole competence that we needed to develop. And helping people basically spread the service to their friends. That was one of the things. Once we got very good at that, that was one of the things that made me feel like, hey, if Instagram joined us early on, then I felt like we could help grow that quickly. And same with WhatsApp. And that's sort of been a core competence that we've developed and been able to.”

    2023-06-08 · Lex Fridman Podcast · #383 – Mark Zuckerberg: Future of AI at Meta, Facebook, Instagram, and WhatsApp · IDENTIFIED FROM THE TRANSCRIPT · source

  24. “And that makes it so that conversations can just filter and fork into all these different directions in a way that can be super useful. So I think there's a lot of things that are really awesome about the experience. It just always struck me. I always thought that Twitter should have a billion people using it or whatever the thing is that basically ends up being in that space. And for whatever combination of reasons, again, these companies are complex organisms and it's very hard to diagnose this stuff from the outside.”

    2023-06-08 · Lex Fridman Podcast · #383 – Mark Zuckerberg: Future of AI at Meta, Facebook, Instagram, and WhatsApp · IDENTIFIED FROM THE TRANSCRIPT · source

  25. “Well, I think text is very accessible for people to transmit ideas and to have back and forth exchanges. So it, I think, ends up being a good format for discussion in a lot of ways uniquely good. If you look at some of the other formats or other networks that have focused on one type of content like TikTok is obviously huge, right? And there are comments on TikTok. But I think the architecture of the service is very clearly that you have the video as the primary thing and there's comments after that. But I think one of the unique pieces of having text-based comments, like content, is that the comments can also be first class.”

    2023-06-08 · Lex Fridman Podcast · #383 – Mark Zuckerberg: Future of AI at Meta, Facebook, Instagram, and WhatsApp · IDENTIFIED FROM THE TRANSCRIPT · source

  26. “And look, I don't know exactly how this is going to turn out. I mean, what I can say is, yeah, there's some people working on this, right? I think that there's something there that's interesting to explore”

    2023-06-08 · Lex Fridman Podcast · #383 – Mark Zuckerberg: Future of AI at Meta, Facebook, Instagram, and WhatsApp · IDENTIFIED FROM THE TRANSCRIPT · source

  27. “I think that it's possible that something that meld some of those ideas with the graph and identity system that people have already cultivated on Instagram could be a kind of very welcome contribution to that space. But I don't know, we work on a lot of things all the time, though, too. So I don't want to get ahead of myself. And we have projects that explore a lot of different things. This is certainly one that I think could be interesting.”

    2023-06-08 · Lex Fridman Podcast · #383 – Mark Zuckerberg: Future of AI at Meta, Facebook, Instagram, and WhatsApp · IDENTIFIED FROM THE TRANSCRIPT · source

  28. “Is a project. I've always thought that sort of a text based kind of information utility Just a really important thing to society. And for whatever reason, I feel like Twitter has not lived up to what I would have thought its full potential should be. I think that the current, you know, I think Elon thinks that, right? And that's probably one of the reasons why you bought it. I do think there are ways to consider alternative approaches to this. And one that I think is potentially interesting is this open and federated approach where you're seeing with Mastodon and you're seeing that a little bit with blue sky.”

    2023-06-08 · Lex Fridman Podcast · #383 – Mark Zuckerberg: Future of AI at Meta, Facebook, Instagram, and WhatsApp · IDENTIFIED FROM THE TRANSCRIPT · source

  29. “Like, maybe the person didn't know, and I'm just going to help educate them, or you could kind of come at it as no, I need to shut this thing down immediately. It's like, I just am not going to talk about this. There may be times where you I actually think having a somewhat more informative approach where you generally assume good intent from people is probably a better balance to be on as many things as you can be. You're not going to be able to do that for everything. But you're kind of asking about how I approach this and I'm thinking about this as it relates to. To AI. And I think that that's a big difference in kind of how to handle sensitive content across these different modes.”

    2023-06-08 · Lex Fridman Podcast · #383 – Mark Zuckerberg: Future of AI at Meta, Facebook, Instagram, and WhatsApp · IDENTIFIED FROM THE TRANSCRIPT · source

  30. “One proposed response is like, no, I can't talk about that, right? It's like basically just like shut it down immediately. Which I think is some of what you see. It's like as a large language model, I'm not allowed to talk about whatever. But there's another response which is like, hey, I don't think that's a good idea. And a lot of countries, including the US, 3D printing guns is illegal or kind of whatever the factual thing is. And I was like, okay, you know, that's actually a respectful and informative answer. And I may have not known that specific thing. So there are different ways to handle this that I think kind of you can either assume good intent”

    2023-06-08 · Lex Fridman Podcast · #383 – Mark Zuckerberg: Future of AI at Meta, Facebook, Instagram, and WhatsApp · IDENTIFIED FROM THE TRANSCRIPT · source

  31. “It's not just looking at a specific example of speech that it's most effective to handle this stuff. And I think often you don't want to make specific binary decisions of kind of this is allowed or this isn't. I mean, we talked about fact-checking or Twitter's community voices thing. That's another good example. It's like it's not a question of is this allowed or not? It's just a question of adding more context to the thing. I think that that's helpful. So in the context of AI, which is what you were asking about, I think there are lots of ways that An AI can be helpful. With an AI, it's less about censorship, right? Because, and it's more about what is the most productive answer to a question? One case study that I was reviewing with the team is someone asked, Can you explain to me how to 3D print a gun”

    2023-06-08 · Lex Fridman Podcast · #383 – Mark Zuckerberg: Future of AI at Meta, Facebook, Instagram, and WhatsApp · IDENTIFIED FROM THE TRANSCRIPT · source

  32. “True, absolutely. Yeah, no, I think you want to be careful about that. I'm not sure I'm expressing this very, very clearly because I certainly agree with your point there. And my point isn't that we should. Not have people on our services that are being controversial. That's certainly not what I mean to say. It's that often, I think,”

    2023-06-08 · Lex Fridman Podcast · #383 – Mark Zuckerberg: Future of AI at Meta, Facebook, Instagram, and WhatsApp · IDENTIFIED FROM THE TRANSCRIPT · source

  33. “Just active disagreements in society where some people may think that something is true or false, the other side might think it's The opposite or just unsettled, right? And those are some of the most difficult to kind of handle, like we've talked about. One of the lessons that I feel like I've learned is that A lot of times You can The best way to handle this stuff more practically is not in terms of answering the question of should this be allowed, but just like. Is the best way to deal with someone being a jerk? Is the person basically just having repeat behavior of Causing a lot of issues. So looking at it more at that level.”

    2023-06-08 · Lex Fridman Podcast · #383 – Mark Zuckerberg: Future of AI at Meta, Facebook, Instagram, and WhatsApp · IDENTIFIED FROM THE TRANSCRIPT · source

  34. “Yeah, I mean, I've probably gotten a somewhat more nuanced view just because I think that there are, you know, I come at this, I'm obviously very pro freedom of expression. I don't think you build a service like this that gives people tools to express themselves unless you think that people expressing themselves at scale is a good thing, right? So I didn't get into this to try to prevent people from expressing anything. I like want to give people tools so they can express as much as possible. And then I think it's become clear that There are certain categories of things that we've talked about that I think almost everyone accepts are bad and that no one wants and that are illegal even in countries like the US where you have the First Amendment that's very protective of enabling speech. It's like you're still not allowed to do things that are going to immediately incite violence or violate people's intellectual property or things like that. So through those, but then there's also a very active core of.”

    2023-06-08 · Lex Fridman Podcast · #383 – Mark Zuckerberg: Future of AI at Meta, Facebook, Instagram, and WhatsApp · IDENTIFIED FROM THE TRANSCRIPT · source

  35. “I don't know. This is one of the most pressurized debates, I think, in society. So I just think that there are so many people in different forces that are trying to apply pressure from different sides that you can make decisions based on trying to make people happy. I think you just have to Do what you think is the right balance and accept that people are going to be upset no matter where you come out on that.”

    2023-06-08 · Lex Fridman Podcast · #383 – Mark Zuckerberg: Future of AI at Meta, Facebook, Instagram, and WhatsApp · IDENTIFIED FROM THE TRANSCRIPT · source

  36. “For a lot of these things, they haven't yet been hardened into a single truth. Society's sort of trying to hash out what we think on certain issues. Maybe in a few hundred years, everyone will look back and say, hey, no, it wasn't it obvious that it should have been this. But, you know, no, we're kind of in that meat grinder now. and working through that. So, no, these are all very complicated. Some people raise concerns in good faith and just say, hey, this is something that I want to flag for you to think about. Certain people, I certainly think come at things with somewhat of a more kind of punitive or vengeful view of like I want you to do this thing. If you don't, then I'm going to try to make your life difficult in a lot of other ways.”

    2023-06-08 · Lex Fridman Podcast · #383 – Mark Zuckerberg: Future of AI at Meta, Facebook, Instagram, and WhatsApp · IDENTIFIED FROM THE TRANSCRIPT · source

  37. “Is there so much pressure from all sides that I'm not sure that any specific thing that someone says is really adding that much more to the mix. There are obviously a lot of people who think that we should be censoring more content. Or there are a lot of people who think we should be censoring less content. There are, as you say, all kinds of different groups that are involved in these debates, right? So there's the kind of elected officials and politicians themselves, there's the agencies. But there's the media, there's activist groups, this is not a US specific thing. There are groups all over the world and kind of all in every country that bring different values. So it's just a very active debate. And I understand it, right? I mean, these are these kind of questions get to really some of the most important social debates that are being had. So it gets back to the question of truth.”

    2023-06-08 · Lex Fridman Podcast · #383 – Mark Zuckerberg: Future of AI at Meta, Facebook, Instagram, and WhatsApp · IDENTIFIED FROM THE TRANSCRIPT · source

  38. “I've seen it. It's really hard from the outside to know exactly what happened in each of these cases. We've obviously been. In a bunch of our own cases where Agencies or different folks will just say, hey. Here's a threat that we're aware of. You should be aware of this too. It's not really pressure as much as it is just flagging something that our security systems should be on alert about. I get how some people could think of it as that. But at the end of the day, it's our call on how to handle that. But I mean, I just, you know, in terms of running these services, want to have access to as much information about what people think that adversaries might be trying to do as possible.”

    2023-06-08 · Lex Fridman Podcast · #383 – Mark Zuckerberg: Future of AI at Meta, Facebook, Instagram, and WhatsApp · IDENTIFIED FROM THE TRANSCRIPT · source

  39. “That I think is a there are more harms that I think can be associated with that. And I think that that ends up being a more critical thing to defend against governments than whereas if another government has a different view of what should be acceptable speech in their country, especially if it's a democratically elected government and I think that there's a certain amount of deference that you should have to that.”

    2023-06-08 · Lex Fridman Podcast · #383 – Mark Zuckerberg: Future of AI at Meta, Facebook, Instagram, and WhatsApp · IDENTIFIED FROM THE TRANSCRIPT · source

  40. “By them, but for whatever reason, we just feel like, hey, this government does not have a good track record. Basically, not trying to get access to people's data. And at the end of the day, I mean, if you put a data center in a country and the government wants to get access to people's data, then they do at the end of the day have the option of having people show up with guns and taking it by force. So I think that there's a lot of decisions that go into how you architect the systems years in advance of these actual Confrontations that end up being really important.”

    2023-06-08 · Lex Fridman Podcast · #383 – Mark Zuckerberg: Future of AI at Meta, Facebook, Instagram, and WhatsApp · IDENTIFIED FROM THE TRANSCRIPT · source

  41. “Sense and freedom of expression at some level. But a government getting access to data in a way that seems like it would be unlawful in our country exposes people to real physical harm. And that's something that in general we take very seriously. And then that flows through all of our policies in a lot of ways, right? By the time you're actually litigating with a government or pushing back on them, that's pretty late in the funnel. I'd say a bunch of this stuff starts a lot higher up in the decision of where do we put data centers. And there are a lot of countries where we may have a lot of people using the service in a place. It might be good for the service in some ways, good for those people if we could reduce the latency by having a data center near”

    2023-06-08 · Lex Fridman Podcast · #383 – Mark Zuckerberg: Future of AI at Meta, Facebook, Instagram, and WhatsApp · IDENTIFIED FROM THE TRANSCRIPT · source

  42. “Person who's leading it is just trying to push in a certain direction. These are very complex questions, but I think it's difficult to have one size fits all approach to it. But in general, we're pretty active in kind of advocating and pushing back on. Requests to take things down. But honestly, the thing that I think a request to censor things is one thing, and that's obviously bad. But where we draw a much harder line is on request for access to information, right? Because if you can get told that you can't say something, I mean, that's bad, right? I mean, that is, obviously it violates your...”

    2023-06-08 · Lex Fridman Podcast · #383 – Mark Zuckerberg: Future of AI at Meta, Facebook, Instagram, and WhatsApp · IDENTIFIED FROM THE TRANSCRIPT · source

  43. “I think we have a bunch of principles around how we deal with that. Because on the one hand, if there's a democratically elected government and people around the world just have different values in different places, then should we as California-based company tell them that something that they have decided is unacceptable actually like that we need to be able to express that? I mean, I think that that's there's a certain amount of hubris in that. But then I think there are other cases where it's like a little more autocratic and you have the dictator leader who's just trying to crack down on dissent and the people in a country are really not aligned with that. And it's not necessarily against their culture, but the”

    2023-06-08 · Lex Fridman Podcast · #383 – Mark Zuckerberg: Future of AI at Meta, Facebook, Instagram, and WhatsApp · IDENTIFIED FROM THE TRANSCRIPT · source

  44. “I don't know that there's a one size fits all answer to that. I mean, I think we basically have the principles around we want to allow people to express as much as possible. We have developed clear categories of things that we think are. Wrong that we don't want on our services and we build tools to try to moderate those. So then the question is okay, what do you do when a government says, That they don't want something On the service”

    2023-06-08 · Lex Fridman Podcast · #383 – Mark Zuckerberg: Future of AI at Meta, Facebook, Instagram, and WhatsApp · IDENTIFIED FROM THE TRANSCRIPT · source

  45. “But look, I mean, this is really difficult stuff. I think it's really hard to know where to draw the line on what is fact and what is opinion because the nature of science is that nothing is ever 100% known for certain. You can disprove certain things, but you're constantly testing new hypotheses and scrutinizing frameworks that have been long held. Every once in a while you throw out something that was working for a very long period of time and it's very difficult. But I think that just because it's very hard and just because there are edge cases doesn't mean that you should not try to give people what they're looking for as well.”

    2023-06-08 · Lex Fridman Podcast · #383 – Mark Zuckerberg: Future of AI at Meta, Facebook, Instagram, and WhatsApp · IDENTIFIED FROM THE TRANSCRIPT · source

  46. “I think It's very difficult to. Just abstain. But I think we should be clear about which of these things are actual safety concerns and which ones are a matter of preference in terms of how people want information flagged. So we did recently introduce something that allows people Have fact checking not affect the distribution of what shows them their product. So, okay, a bunch of people don't trust you the fact checkers are. All right, well, you can turn that off if you want. But if the content violates some policy,”

    2023-06-08 · Lex Fridman Podcast · #383 – Mark Zuckerberg: Future of AI at Meta, Facebook, Instagram, and WhatsApp · IDENTIFIED FROM THE TRANSCRIPT · source

  47. “So, I do think that the questions around how to manage that are very nuanced. The way that I try to think about it is that I think it's best to generally boil things down to the harms that people agree on. So when you think about is something misinformation or not, I think often the more salient bit is, is this going to potentially lead to physical harm for someone and kind of think about it in that sense? And then beyond that, I think people just have different preferences on how they want things to be flagged for them. I think a bunch of people would prefer to kind of have a flag on something that says, hey, a fact checker thinks that this might be false. I think Twitter's community notes implementation is quite good on this. But again, it's the same type of thing. It's like just kind of discretionarily adding a flag because it makes the user experience better, but it's not trying to take down the information or not. I think that you want to reserve the kind of censorship of content to things that are”

    2023-06-08 · Lex Fridman Podcast · #383 – Mark Zuckerberg: Future of AI at Meta, Facebook, Instagram, and WhatsApp · IDENTIFIED FROM THE TRANSCRIPT · source

  48. “Sense like, all right, are you going to censor someone for just being wrong? If there's no kind of harm implication of what they're doing, I think that that's a bunch of real kind of issues and challenges there. But then there are other places where it is just takes some of the stuff around COVID earlier on in the pandemic where there were real health implications, but there hadn't been time to fully vet a bunch of the scientific assumptions. you know, unfortunately, I think a lot of the kind of establishment on that kind of waffled on a bunch of facts and asked for a bunch of things to be censored that in retrospect ended up being more debatable or true. And that stuff is really tough, right? And really undermines trust in that.”

    2023-06-08 · Lex Fridman Podcast · #383 – Mark Zuckerberg: Future of AI at Meta, Facebook, Instagram, and WhatsApp · IDENTIFIED FROM THE TRANSCRIPT · source

  49. “I really agree with what you're pushing on. I mean, the core The core shape of the problem is that there are some harms that I think everyone agrees are bad, right? So sexual exploitation of children, right? Like you're not going to get many people who think that that type of thing should be allowed on any service. And that's something that we face and try to push off the as much as possible today, terrorism, inciting violence, right? It's like we went through a bunch of these types of harms before. But then I do think that you get to a set of harms where there is more social debate around it. So misinformation, I think, is Has been a really tricky one because there are things that are Kind of obviously false that are maybe factual. May not be harmful”

    2023-06-08 · Lex Fridman Podcast · #383 – Mark Zuckerberg: Future of AI at Meta, Facebook, Instagram, and WhatsApp · IDENTIFIED FROM THE TRANSCRIPT · source

  50. “For these, what we see is for nation states or these actors that are trying to pull off these large coordinated and authentic networks to kind of influence different things. At some point when we would just make it very difficult, they do just try to use other services instead, right? It's just like if you can make it more expensive for them to do it on your service than kind of people go elsewhere. And I think that that's the bar, right? It's like it's not like, okay, are you ever going to be perfect at finding every adversary who tries to attack you? It's, I mean, you try to get as close to that as possible. But I think really kind of economically we're just trying to do is make it so that it's just inefficient for them to go after that.”

    2023-06-08 · Lex Fridman Podcast · #383 – Mark Zuckerberg: Future of AI at Meta, Facebook, Instagram, and WhatsApp · IDENTIFIED FROM THE TRANSCRIPT · source