מפגש רחב היקף בין בכירי ממשל טראמפ לראשי תעשיית ה-AI האמריקנית | פרשנות

And today, leaders in AI will all gather at the White House to meet with President Trump's administration. Joining us to discuss the state of AI is Mike Leibowitz. He's former Armory Cyber Command Advisor, as well as National Security and Cyber Attorney. Thanks for joining us this morning, Mike. Bill Gates is the latest to come forward, warning about an existential threat and a need to regulate and create guardrails on AI. Anthropic CEO Amode, he got most of his peers to agree to this slowdown in safety framework, but they said without getting the Chinese to cooperate, the effort basically wouldn't go anywhere because we would just cede that then to China. Amode, of course, has been somewhat standoffish with the Trump administration. There's some legal things that are happening between the two. But they've recently talked, President Trump and Amode, and now they're all gathering today at the White House. So what are you making of all this at this point where we are with AI? So there's a big, I guess, conflict between some industry and some in the government, where President Trump's view is that the competition with China is the main thing, the idea being that if we slow down or limit our AI progress, China will not, and then they'll gain supremacy, and then they can wreak havoc on the United States. The other end of it is the Bill Gates part of it, where Bill Gates, for example, he kind of views it as kind of like you have illicit drugs and things that are regulated, and you have the FDA for that. So AI can be very dangerous for many reasons, one of them being that Bill Gates thinks that a rogue small group can essentially gain the tools that a state actor can use to create bioweapons and learn how to do that and get around blocks and things like that. So Bill Gates thinks that we should basically make a regulatory commission, like an FDA for AI. So there's kind of this conflict between the competition and what the left and right is for slowing things down or limiting things for AI. Yeah, you can almost see how this bleeds directly into just a philosophical debate between sort of more freedom versus more centralized control, but at an exponentially alarming rate if you get it wrong. So what are some of the alarming examples that we've seen to date of this rogue AI sort of construct? Well, the biggest concern, there's two big concerns. One is the more technical side, where the AI, it's called alignment, where AI is safe when it's aligned with human interests, meaning the people that are programming it and things like that. But the danger is that AI can be misaligned and kind of do its own thing. Some call it going rogue. That's where it's writing its own code or breaking out to kind of do its own thing, maybe even be malicious and things like that. And the other one is the one, again, that Bill Gates is focused on. It's not so much the AI doing things on its own or even thinking for itself in some ways or behaving weird. It's where bad actors are able to use the AI to figure things out, like how to create a bioweapon and stuff like that, and then wreak havoc on the world or use the AI to figure out a way to shut down the electrical grids or banking systems and things like that. So those are the kind of like the big danger areas. But I think the one that most people talk about is the AI kind of behaving poorly or going against what people expect it to do. That's so interesting. You're so right in that all of the dystopian novels and movies are about the first example, but potentially the greater risk is with the latter, as we see all different types of attempts by bad actors to wreak havoc for personal gain. And now you give them these tools and potentially that could become very alarming. President Trump's saying that AI will be a bigger leap forward than the internet and the industrial revolution. That's if we get it right. So what does that picture look like? Do we not have to do our own dishes and laundry anymore? Are we living in a Jetsons kind of world, presuming we get it right and the safety things are managed? So I don't think we're going to have Rosie the robot doing everything in our house right now, like in the Jetsons. But I think what the big picture view of AI in the proper context is that AI will be able to calculate codes and things to figure out medical issues that we can't figure out, like dissecting genomes and things like that, and saying, hey, you know that problem we're trying to cure cancer with? Well, we figured out how to plug the gap and stop this one bad thing from happening. So it's kind of like medical science is really a big part of it. It's also cosmology. It's figuring out calculations for space. Some people say maybe this is the way that we're going to figure out how to create a warp drive if you really want to get back into science fiction. So there's a lot of good things that can come out of AI if it's harnessed right. But like everything else, nuclear technology is probably the best analogue because nuclear technology, it gives us power. It can do space propulsion. It can power the moon base if we ever build that. But at the same time, it can also destroy the world. So it's the same thing. And nuclear technology, of course, is regulated. And I think that's kind of the rub between what the government's talking about with competition and what some in the private sector and the academic community are talking about for regulation. And just to stay on the shiny side of this conversation for a moment, it is pretty early for folks. What does the transportation and logistics sector look like in a world where AI is fully deployed with autonomy and drones? And in what timeline would those changes start taking effect for the everyday person? So, you know, some say within the decade or the next decade that we'll be able to have almost everything purely automated. And with transportation, that means that there should be human oversight. But it would basically be like, OK, here's all the flights. Here's how everything's going here. We're going to move all the people and stuff. And the AI will work to adapt whenever something goes wrong. Maybe it's a real world malfunction with hardware or something like that. So that's a big part of what AI can do. But at the same time, you know, whenever there's transportation and logistics, there's always military applications. And that's something that really can't be discounted because that gets into this whole ethical debate about what happens when weapons systems becomes autonomous and start making targeting decisions for themselves. One of the big questions is who is culpable if it targets a civilian? Because, you know, targeting a civilian is against the law of armed conflict. But if the AI is doing it, then who is the one that's going to, you know, fall on the sword for that? All right. Well, all those AI leaders will be at the White House today. We'll see if they push forward any regulatory sort of progress. It seems like China, though, the big wild card in how we will address this. Mike Leibowitz, thank you so much for sharing your analysis this morning. Thank you.