90 views
•
6 years ago
0
0
Share
Save
10 appearances
Lex Fridman is a computer scientist and researcher in the field of artificial intelligence and robotics, and host of the Lex Fridman Podcast. https://youtube.com/lexfridman
45 views
•
6 years ago
92 views
•
6 years ago
37 views
•
6 years ago
Show all
No, is there, I don't know anything about coding, but do you have like, is there a spell check for coding? Yeah. So it's kind of called debugging. It's trying to find bugs. And it's a software that's doing this? Yeah, software. So there's, depending on the programming language and everybody should, uh, if you, uh, if you haven't tried programming, you should try it. It's, it's cool. It's the future. You should learn to program. Okay. That's my plug. You said, well, say learn to code. You can get kicked off. What did that, I heard that everyone's scared of it. I it's a problematic term. I don't actually know why it's the dumbest fucking problematic code of all time, because someone ridiculously was suggesting that coal miners could maybe learn how to code computer code and like get a different job. They could be trained. And so as someone, the way people were looking at it, like that, that was a, uh, like a frivolous suggestion and that it was ridiculous to try to get someone who was 50 years old. It doesn't have any education in computers at all to change their job from being a coal miner to learning how to code. So they started saying it to politicians and people mocking it. But then what Twitter alleged was that what was going on was it was being connected to white supremacy and antisemitism and a bunch of different things like people were saying, learn to code. And they were putting in a bunch of these other phrases in my suggestion would be, well, that's a different fucking thing. Like now you have, like you look, you have a problem with Nazis and white supremacists, but that's the problem is with Nazis and white supremacists. When someone is just saying, learn to code, mocking this ridiculous idea that you're going to teach, you know, that's a, that's a legitimate criticism of someone's perspective that you're going to get a coal miner to learn how to fucking do computer coding. It's crazy. It's a, it's a, so people getting banned for that. Rightly so people were furious. The way Google described it to me and Tim Poole and we were discussing it was that Google, I mean, excuse me, Twitter, the way Twitter described it was that essentially they were dealing with something where they were trying to censor things at scale. There was, there was so many people and there's so much going on that it's very difficult to get it right. And that they've made mistakes. I think that's a fast, one of the most fascinating applications of AI actually is filtering, trying to manage. Computer learning. The little, so using machine learning to manage this huge conversation. You talk about 500, I believe it's 500 million tweets a day, something like that. And he Jamie makes at least three. Three. That might be one. I was going to say with this conversation, I saw this, uh, recently, I don't know who did the data on this, but there's a, uh, a, a statement someone put on Twitter that said that of, um, let me see if I can word it correctly. It was 22% of adult Americans are on Twitter. Whoa. All right. So that's like, that's like a fact one of that. 10% make up 80% of the tweets created by adult Americans. So that's 2% of the people on Twitter make up 80% of the tweets. Yeah, that makes sense. Yeah. A lot of people are arguing. Aggressively. And the question on how to manage that and you can't manage that, but just, uh, manual, uh, review of each individual tweet. Yeah. You'd have to have so many employees. Yeah. That's, I think more likely. I don't think Jack is lying, but, um, nor is Vija, but I do think that they have a clear bias against conservatives and that's being shown. So that's an interesting question. I have, uh, your friend, my friend and mentor Eric Weinstein, you talked to me. I do disagree with him a little bit on this. I think, uh, he basically believes, so there's a bias, it boils down to the conversation that Jack is having at the, at the top level inside Twitter, what, what is that conversation like? Uh, I think I tend to believe again, this might be my naive nature is that they have, they don't have bias and they have just, they're trying to manage this in huge flood of, um, of tweets. And what they're trying to do is not by, is not to remove sort of conservatives or liberals and so on. They're trying to, uh, remove people that lead to, uh, others leaving the conversation. So they want more people to be in the conversation. I think that's true as well, but I think they definitely are biased against conservative people. There was a, an Alexander, Alexandra, AOC Octavia. How is it? AOC is good. Cortez is the last one. Is it Octavia? Ocasio. That's right. Okay. I'm sorry. Alexander, AOC. Sorry. I'm just, I'm thinking I wasn't, wasn't planning on talking about her, but, um, there was a parody account and someone was running this parody account, which was very mild, just humorous parody account. They were banned permanently for running it. And then their own account was banned as well. Whereas, um, you know, there's some progressive people or liberal people that post all sorts of crazy shit and they don't, they don't get banned at the same rate. It's really clear that someone in the company, whether it's up for manual review, whether it's at the discretion of the people that are employees, when you're thinking about a company, that's a Silicon Valley company, you are in, without doubt, you're dealing with people that are leaning left. There's so many that lean left in Silicon Valley. The idea that that company was secretly run by Republicans is ridiculous. They're, they're almost all run by Democrats or progressive people. So at the leadership level, there's, there's a narrow-mindedness that, that, that, that permeates all of Silicon Valley. You're saying, well, the question is, I think there's a leaning left that permeates Silicon Valley. I think that's undeniable. I think it's undeniable. I mean, I think if you had a poll, the people that work in Silicon Valley, where their political leanings are, I think it would be by far left. I think it would be the vast majority. Um, does that mean that affects their decisions? Well, what's the evidence? Well, it's kind of shows that it does, you know, it's, they're not treating it with a hundred percent clarity and, and, you know, across the board accuracy or, um, fairness rather, I think that there's absolutely people that work there that lean. And there's been videos where they've captured people, uh, that were Twitter employees talking about it, talking about how you do that, how you, uh, make their, you know, find, you find someone who's using Trump talk or, you know, saying sad at the end of things. And someone who's talking, he's going to, you know, that certain characteristics they look for, and there's been videos of what is that project Veritas where that guy, uh, got his employees got undercover footage of Twitter employees talking about that kind of stuff. Question is how much power do those individuals have? How many individuals are there like that? Are that, are those people exaggerating their ability and what they do at work or they, are they talking about something that used to go on, but doesn't go on anymore? I don't know. I don't work there. I think it boils down to, I, I'm one of those people that believes it's, it boils down to the leadership. Some people at the top set the culture and the culture has to be, uh, it cannot be this kind of Silicon Valley, narrow-minded, uh, sort of left leaning thinking. Even if you believe, even if you're a hardcore liberal, you cannot, when you operate the com, when you drive and manage the conversation in the entire world, you have to think about middle America. You have to think about, you have to have fundamental respect for human beings who voted for Trump. It is a concerning thing for me to see just a narrow-mindedness of, in all forms. One of the reasons I enjoy listening to this podcast is you're pretty open-minded. That open-mindedness is essential for leaders of Facebook and Twitter, people who are managing conversations. I think so too. I think it's, I think it's a, the thought of being open-minded and, and acting in that ethic is probably one of the most important things that we could go forward with right now because things are getting so greasy. It's so slippery on, on both sides. And we're at this weird position that I don't recall ever in my life. There being such a divide between the right and the left in this country. I don't, it's more, more vicious, more angry, more hateful. It's different than at any other time in my life. And I think a lot of our ideas are based on these narratives that may or may not even be accurate. And then we support them and we reinforce them on either side. We reinforce them on the left. We reinforce them on the right. Where if you're looking at reality itself and you don't have these clear parameters and these clear ideologies, I think we're way, most of us are way more in the middle than we think we are. Most of us are. We just don't want racists running the country. We don't want socialists giving all our money away. We don't want to pay too much in taxes to a shitty government. We don't want schools getting underfunded. We all, you know, and then we decide what, what does my team like the team that I, the shit that I like, is that this team? Well, not everything, but they got a lot of things, so I'll go with them. Maybe I'm not a religious nut, but I'm fiscally conservative and I don't like what Democrats like to spend money. I'm going to go with the Republicans. You know, maybe, maybe I'm, you know, more, I'm more concerned with the state of the economy and the way we trade with the world than I am with certain social issues that the Democrats embrace. So I'll lean that way, even though I do support gay rights and I do support this and I do support all these other progressive ideas. There's way more of us in that boat. There's way more of us that are in this middle of the whole thing. For sure. But there, it goes up and down. So all of us, so I'm, I believe, I hope I am open-minded most of the time, but you have different moods. Oh, for sure. Yeah. And the question is, this is where the role of AI comes in. Does the AI that recommends what tweets I should see, what Facebook messages I should see, is that encouraging the darker parts of me or the, the Steven Pink or better angels of our nature? Like, is it, what stuff is it showing me? Because if it shows me, uh, stuff that, like if the AI trains purely on clicks, it may start to learn when I'm in a bad mood and point me to things that might be upsetting to me. Yeah. And so escalating that division and escalating this vile thing that can be solved most likely with people training a little more jiu-jitsu or something. Well, this Facebook algorithm that encourages people to be outraged because accidentally, not even on purpose, but this is what engages people. This is what gets clicks. So they find out, oh, well, he clicks on things when he finds out the people are anti-vaccination or he clicks on things when he finds out, you know, whatever, fill in the blank with whatever the subject is. And then you get these motherfuckers, you know, there's a reason why measles is spreading and you start getting angry. I mean, the anti-vax arguments on Facebook, I don't know if you ever dip into those waters for a few minutes and watch people fight back and forth and in fury and anger, you know, it's, uh, it's another one of those things that becomes a extremely lucrative, uh, subject for any social media empire. If you're, if you're all about getting people to engage and that's where the money is in advertising, yeah, to getting people to click on the page and the ads are on those page, you get those clicks, get that money. If that's how the system is set up, and I'm not exactly sure how it is because I really use Facebook, but that's what it benefits. I mean, that's what, that's what it gravitates towards, gravitates towards controversy. So, and when we think about concern for AI systems, we'll talk about sort of Terminator, I'm sure we'll, we'll touch on it, but I think of Twitter as a whole as one organism. That is the thing that worries me the most is the artificial intelligence that is very kind of dumb and simple, simple algorithms that are driving the behavior of millions of people. And together, the kind of chaos that we can achieve, I mean, that algorithm has incredible influence on all society. Twitter, our, our current president is on Twitter. So much, yeah, all day, all night. The, the, I mean, it's scary to think about, we talk about autonomous vehicles leading to fatal, one fatality, two fatalities. It's scary to think about what the difference between a small change in the, in the Twitter algorithm, I mean, I, it could start wars. It really could. And that, if you think about the long term, if you think about as one AI organism, that is a super intelligent organism that we have no control over. And I think it all boils down honestly to the leadership, to Jack, to, to, to, and other folks like him, making sure that he's open-minded. It goes hunting, that he goes, does some jiu-jitsu, that he eats some meat and sometimes goes vegan. Right?