Godfather of AI Warns: AI Could Become Uncontrollable

In the near future, AI may be used to create terrible new viruses and horrendous lethal weapons that decide by themselves who to kill or maim. We urgently need research on how to prevent these new beings from wanting to take control. They are no longer science fiction. Welcome back to another episode of Silicon Valley Front Row. I'm David Lam and I'm with Steve Ispas. So Steve, today, what are we talking about? We're talking about AI. Again? So how come we always talk about AI? Well, it seems like it's one of the most important things that happen, not just in the Silicon Valley, but around the world. And we talked to a lot of experts. We got a lot of feedback, a lot of expertise about AI and how it's influencing. You know, if you go on Highway 101 from San Jose to San Francisco, you see all these ads, all these billboards. Almost every single one is about AI. You know, AI here, AI there. So it's a big thing. So I think we have an interesting topic. Yeah, it's an industry or you could even say a world disruptor. It changes every aspect of life, pretty much. So because that's an important topic, we thought you'd play some of what George Hinton has to say on this topic. And George Hinton is a computer scientist who's actually known as the godfather of AI. That's right. And I think that's because he worked on artificial neural networks. And at least from what I understand is that in the brain, we have the biological neural networks. But he designed artificial neural networks that work in computers and obviously compute at a very fast speed. And I think he based the structure and the functions of those artificial neural networks is the same as the human brain. So it's pretty cool stuff. I mean, it's pretty amazing what he did. So I had to look up what neural networks are. So just to give like a definition to our viewers. So a neural network is a machine learning model that stacks simple neurons in layers and learns pattern recognizing weights and biases from data to map inputs to outputs. There's the computer part of it, right? That artificial one. Right. So I would assume the brain does the same thing in some ways. They copy that. But that's pretty interesting. So anyway, let's kind of see what he has to say. And then we'll stop it and comment. Your Majesties, Your Royal Highnesses, Excellencies, Dear Laureates, Ladies and Gentlemen. This year, the Nobel Committees in Physics and Chemistry have recognized the dramatic progress being made in a new form of artificial intelligence that uses artificial neural networks to learn how to solve difficult computational problems. And again, that's the stuff that he invented, the artificial neural net. So that's why I think he's so familiar with it. This new form of AI excels at modeling human intuition rather than human reasoning. And it will enable us to create highly intelligent and knowledgeable assistants who will increase productivity in almost all industries. If the benefits of the increased productivity can be shared equally, it will be a wonderful advance for all humanity. Unfortunately, the rapid progress in AI comes with many short-term risks. It has already created divisive echo chambers by offering people content that makes them indignant. It is already being used by authoritarian governments for massive surveillance. So I thought it'd be important to stop there because we just did an episode that talked about the Cisco case, how Cisco Systems was aiding the Chinese Communist Party in following and monitoring Falun Gong practitioners. And so I'm pretty sure when he talks about that, he's referring to China. It could be other regimes too, but China is probably the main one he's referring to. So do you remember we talked about this? Yeah, it was the Cisco CCP surveillance episode. And it was based on a project called the Golden Shield. Yes. So their software apparently helped Chinese Communist Party monitor people. Track and monitor, yeah. Based on keywords, right? Yes. And this was advertised particularly for CCP officials looking for Falun Gong. Yeah. And you can only imagine if there was 12 years ago or 13 years ago, the lawsuit was filed, how much more can be done now with AI? Maybe we should read some of these comments here that we came across. So this is from former FBI Director Christopher Wray. He said, China is engaged in a highly sophisticated malign foreign influence campaign, and its methods include bribery, blackmail, and covert deals. We've now reached the point where the FBI is opening a new China-related counterintelligence case about every 10 hours. So I think that the whole thing with AI is that if it's used for good things, it can benefit people in whatever way. But the bad people can use it in a way that makes it much more complicated and harder to track. Right. And let me read another comment on the same topic. China conducts the most sophisticated global and comprehensive campaign of transnational repression in the world. The CCP targets entire ethnic and religious groups, including Falun Gong practitioners. So anyways, let's keep going. I thought this was very relevant, what he was talking about. And by cybercriminals for phishing attacks. In the near future, AI may be used to create terrible new viruses and horrendous lethal weapons that decide by themselves who to kill or maim. So I think also what's interesting is when AI moves towards what is called the super AI, the artificial super intelligence, that's when these things become much more possible as far as they can create, you know, someone can just give a command or by themselves they can create a virus that's, like you said, targets different people versus... Basically, there's no limit of how advanced the technology can get once AI gets advanced as well. So let's see what he says. All of these short term risks require urgent and forceful attention from governments and international organizations. There is also a longer term existential threat that will arise when we create digital beings that are more intelligent than ourselves. So I think what he means by digital beings is the same as what's ASI, artificial super intelligence. I think that those terms are pretty much the same thing. So that's what it means by digital beings. We have no idea whether we can stay in control. But we now have evidence that if they are created by companies motivated by short term profits, our safety will not be the top priority. We urgently need research on how to prevent these new beings from wanting to take control. They are no longer science fiction. Thank you. Yeah. So what do you think? I think it's something we definitely need to think about, especially if it's coming from the godfather of AI, a person that helped do a lot of research for artificial intelligence. So I think it's a valid concern. Yeah, well, that's why I thought we'd bring this episode so people get a bit more understanding of the power, and if it's not used right, of how AI can be very damaging. Did we have some questions from the last episode that they wanted to go over? Before we get to that, I kind of want to read... The comments, right? Yeah, the comments and also some of the headlines that's been in the news lately regarding Geoffrey Hinton. So some of the headlines in the news, top stories are like, Godfather of AI issues stark warning to techies who are thinking of changing the stream due to artificial intelligence boom. Another is Godfather of AI predicts total breakdown of society. And on the other side, there's also regulation and people are advocating to have safeguards for AI. For example, the other day we spoke to Samir Kumar, venture capitalist in AI. He said one of the scariest prospects is having a super AI that's able to go past its rules that they shouldn't ever break. Like, for example, do you remember the supposedly the Ten Commandments? You know, it shouldn't hurt humans. That's right, that's right. It should be the number one law for robots. But then we don't know if it could override rules like that not to hurt human. If one of its objective is to say... To create some objectives or to... Yeah, you're right. Yeah, to complete its mission. Because one understand these models can rewrite themselves. If it reaches the stage of ASI, they can rewrite themselves to become optimal or better or more advanced. So when it rewrites itself, then it can probably decide, you know, we don't need this rule. Why? Why is this ruling here to begin with? Right, it's like a double-edged sword. It's able to think and become better. But then what if it thinks it's better than these rules that people have set? Yeah. Yeah, that's pretty crazy. So let's read some comments that we've received from viewers. All right, so here's the first comment. It's from Stay Vigilant and we thank you for your comment. It says, AI trading is so hype nowadays. I'm glad I watched this video and learned what the unbiased research says. Now, of course, we can't verify exactly that the research was unbiased, but it sure seems that way. This is about the 95% of AI implementations failing. Right. By MIT students. Right. So we feel that was pretty unbiased. And yeah, we appreciate your comment. And I think we hope that these studies continue and they're not being suppressed in any way, or again, conducted by some kind of think tank that's being funded by an AI company. Yeah, these things should definitely have independent studies, especially if it's hitting every sector of the world, right? Food, jobs, the economy, to even having AI in our kitchen, the microwaves, phones. So the second comment is from Zoe and she said, good channel. So we've gotten a lot of these comments, but I just thought it'd be nice to give a shout out and recognize one of them. You know, it means a lot, right? When I see these thank yous and things like that, it's really good. So thank you so much, Zoe. So please put your comments below. What do you think about Geoff Hinton's speech as far as the dangers of AI? Please put your comments below and we look forward to review them and read some of them on our next show.