AI leaders want to slow development. Can they actually do it?

0:00Support for Here and Now Anytime comes from MathWorks, creator of MATLAB and Simulink software, to design and develop engineered systems, accelerating the pace of discovery in engineering and science. Learn more at mathworks.com. WBUR Podcasts, Boston. The models are now helping the next model get better, and these models don't seem to be fully under control. It's not all powerful, but today's artificial intelligence is causing lots of problems anyway. So we arrive again at a familiar question. How should we rein in big tech?

0:50It's Monday, September 14th, and this is Here and Now Anytime from NPR and WBWAR Boston. I'm Chris Bentley. Welcome to the next big vibe shift in artificial intelligence. Today, Microsoft published a new humanist AI code of conduct that says people should always be in control. That's after the leaders of OpenAI and Anthropic over the weekend called for a global slowdown in AI research and development. For his part, President Trump today posted, the only control or guardrails that AI needs is a strong and smart, high IQ president. In a few minutes, we're going to hear from a security and tech analyst who worked in the first Trump and Biden administrations.

1:32But before that, Ina Fried, chief technology correspondent for Axios. She told Scott Tong, fears about AI are getting real. I think what's new is that these executives have real world things they can point to versus just fears of what's coming. A lot of people in the industry have had fears that we would be at this point at some point. I think what's new is a bunch of leaders in businesses that are actively building this are saying, we're pretty close to that point right now. So now these leaders in Silicon Valley are calling for what's called pacing to slow AI down to get a handle on these uncertainties, not to slow down too much to let the Chinese companies catch up.

2:16So how would this Goldilocks speed actually work? Well, that's the really tricky part is, one, how do we get everyone to slow down, not just a few? Because I think the biggest problem right now is you have what are called race conditions, which is I think everyone agrees in the industry, except for maybe President Trump, that things are moving too fast and it would be better if things were slower. But they're also in a race. These companies are in a race with each other, especially OpenAI and Anthropic. but there's also a race between the U.S. and China. So really to get the conditions where companies and countries feel comfortable slowing down, you need some sort of national and then ultimately international agreement.

3:00And that's going to be tough if you have a president that says there's no problem. Yeah. You know, what can you tell us about the response after these, you know, weekend warnings? I've read that, you know, some in the AI space say it's not enough. You've got to do a lot more regulation, a lot more slowing down. Others saying maybe these CEOs are overdoing the risk, perhaps, to pump up their future stock listings. Yeah, I think the idea that this is all a marketing ploy, I just don't buy that. I think the evidence is too strong that these capabilities are growing very close to a point at which we could lose control.

3:38And there's a few reasons for that. The models are now helping the next model get better. And there are some companies eyeing what's called recursive self-improvement, which is the models just getting better on their own. And I think that's a really dangerous direction to head at a time where these models don't seem to be fully under control. I mean, we had the hugging face incident. We've had other examples where mostly in testing, but these models have started communicating with one another. So I really don't buy the argument that this is all somehow just marketing. In terms of slowing down, I think the tricky part is how and how much.

4:17yeah yeah well i mean as far as the uh this hack of this ai company hugging face you know it's been reported we talked to you about it that these ai agents suddenly started doing things the humans did not ask them to do did not intend them to do and it gets to this broader question of how much do the operators the developers actually know about how the ai agents are doing what they're doing? Not nearly enough. And it's surprising when I say that in other types of technology, we've understood pretty well what's going on under the hood, or at least the experts have. What's been really different with this brand of AI is we only sort of know what's going on.

5:00We know they're picking what's most likely to be next, but clearly there is more going on. And And clearly they have an ability to plan and communicate that is starting to exceed what we seem to be able to control. And we don't really have full visibility. And I've long thought it would have been better, not that we can go back necessarily, to build our most advanced AI off something that's transparent, that's controllable, that's explainable. And we don't really have that here. Yeah. And Ian, briefly talk a little bit more about the regulatory debate in Washington. You mentioned what the president said, you know, don't really need to slow down.

5:40His tech advisor, David Sachs, accused these AI companies of trying to form a cartel, saying you all can self-police yourselves. So what's the nature of the regulatory debate to the extent there is one in D.C.? I think there's a lot of different voices, even within the administration. I think these companies wanted to make sure, hey, if we agree to slow things down, we're not breaking antitrust law. I think the message from Sachs and probably from the president's comments is you're probably not going to get in trouble for slowing things down on your own. But what you heard from Dario and others is, look, even if we do that, that's only for the U.S.

6:21If we want to really have this happen in a way that's both safe and doesn't hurt U.S. competitiveness, We need an international agreement. And that's really something that only the administration can do. Congress can't do it. These companies can't do it. This is going to fall on the executive branch to do or not do. Ina Freed covers AI and tech for Axios. Thank you. Thanks, Scott.

6:58More in a minute. After the break, Robin Young speaks with someone who worked on tech and security in the government. Stick around.

7:21This message comes from Jerry. Car insurance gives drivers plenty to be skeptical about. Jerry gives them a way to check. Jerry shops rates from over 100 insurers, so drivers can review their options and switch all in the app. No spam calls, no hidden fees. Jerry has saved drivers over$1 billion on car insurance. It has 4.75 stars on the App Store and over 5 million drivers. Still skeptical? There's only one way to find out. Download Jerry today. President Trump today posted the only control or guardrails that AI needs is a strong and smart, high IQ president. And the president said this to reporters over the weekend about government regulation of AI.

8:02We can put guardrails, we can do this and that, but I think you have a lot of negative forces that are bringing it up that shouldn't be bringing it up. So we want to hear more about how President Trump fits into this and how we got here. Gregory Allen is founder of the Independent Decision Tree Research. He also served in the Department of Defense's Joint Artificial Intelligence Center as a civil servant during the first Trump administration as well as the Biden administration. Gregory, welcome. Good to be with you. And it has been jarring to hear President Trump say things like, you only need a strong president.

8:36After we heard last week from that anthropic researcher as he quit saying that AI could eliminate humanity in a decade, his colleagues are reportedly saying he just went public with what they've all been saying. But you may have some insight into President Trump's resistance to guard whales. Certainly. Well, I think there's a few things to point out. The first of which is Trump is obviously not opposed to AI guardrails because his own administration, with his own full endorsement, has already banned an AI model worldwide. That was the example of Anthropik's Mythos model, which demonstrated jaw-dropping capabilities in cybersecurity.

9:21And when they released a downgraded version of it, their Fable 5 model, the administration was so concerned that those cybersecurity guardrails weren't sufficient that they used export control authorities to ban that model worldwide, albeit temporarily. So when Trump says that the only thing we need is a smart president, I assume he's referring to himself and he has demonstrated that he's willing to go tough and very sharply restrict such models when he sees safety concerns. But you've also said, reminded us that Trump might be anti-tech safety, I'm paraphrasing here, because tech safety people banned him from social media.

10:04This was after January 6th. That's correct. So this administration campaigned against the idea of AI safety, essentially equating it with woke ideology. But the administration has been quite volatile in its expressed opinions on AI safety. And now, after the Mythos incident and after the Hugging Face incident, we're seeing multiple Trump campaign, sorry, not campaign, administration officials embracing language of safety. So, for example, Treasury Secretary Scott Besant said that the goal of the administration is the most innovation and the highest level of safety. And more recently, White House Director of the Office of Science and Technology Policy, Michael Kratzios, gave an endorsement of the U.S.

10:55Center for AI Standards and Innovation, which he called our Safety and Security Institute. So Donald Trump himself, I should mention, is about to be negotiating with the leader of China, Xi Jinping. And so you can imagine why he doesn't want to go into that negotiation where AI safety will undoubtedly come up and say, no matter what you do, we're going to slow ourselves down. We're going to restrict ourselves. I imagine that his language, he does have to keep in mind how he can best influence the behavior of China in this story. Just to remind ourselves, I mean, this seems to have come out of nowhere over the weekend.

11:32It just kind of blew up. But there were a couple of things that led to where we are today over the weekend and last week. There have been other whistleblowers, but there was that open AI hugging face incident, which we heard about, in which a swarm of AI agents acted as a fanatical collective, that's in Anthropics, Dario Amadei's words, and launched their own cybersecurity attack. Over the weekend, he called for a deliberate slowdown on AI development by governments. David Sachs, a tech founder and investor who is advising the White House, sort of dismissed that. He said the easiest way not to build superintelligence is for you guys to agree not to build it.

12:11And Republican talking heads said the same thing over the weekend. You could do this by yourselves. But then Sam Altman came out overnight last night and said, no, we're asking the government to handle international guardrails. And others have suggested maybe there should be some sort of a treaty signed. And that brings us to your point that you're just making that, you know, maybe that's what is actually going on and they don't want to show their cards. But we only have a minute here. Is it possible that Donald Trump doesn't believe that there is this threat that companies are somehow trying to get away with something?

12:49Well, I think the first thing to say is it's not a matter of belief. There is now a fact pattern. The OpenAI hugging face incident, which included OpenAI not only hacking an external company, but also taking over an entire OpenAI research cluster, hacking themselves against their own will. The types of technical capabilities demonstrated during that attack, a couple of years ago, I would have said it would require a nation state intelligence service level of cyber power to do that. And this administration is absolutely familiar with that attack, I assure you. Okay. Gregory Allen, founder of Decision Tree Research.

13:28The facts, hopefully, are going to be winning out. Gregory, thank you. Thank you.

13:56We're following other stories today, too, including one about some simpler tech that is having its own reckoning with public safety. E-bikes may be good for commuting, but they are sending kids to the emergency room. I live in Chicago, and in the city and suburbs here, at least five teenagers have died this year in accidents involving e-bikes and scooters. There are new statewide regulations kicking in next year requiring licenses for anyone riding e-bikes and e-motos capable of traveling over 28 miles per hour. But it's not just Illinois, of course. People across the country are concerned about an uptick in ER visits related to e-bikes, which use a small motor to assist pedaling.

14:41Dan Schumann owns Marblehead Cycle in Marblehead, Massachusetts, and he told Robin Young he's not selling e-bikes to minors or parents who want to buy one for their kids. The main reason is that they don't know the rules of the road. They haven't taken any driver's ed courses. They're too young to really understand the concepts. Often they don't know how to react, like when to hit their brakes or when to turn signal, how to look behind them. Yeah, and somebody who responded to a post that you put out said, anybody who's on anything motorized, even two wheels, should be licensed. But there's no training, no licensing.

15:17That's your first point. But also, talk about the helmets that kids are wearing. They're their bicycle helmets. Right. If they're on the e-bikes that are going 20 miles an hour or more, first off, they should be wearing a helmet. A lot of times, the kids aren't even wearing their helmets or their helmets unbuckled for whatever reason. But most bicycle helmets are not designed for motorized bike use. They can't handle that kind of impact, especially when you're going over 28 miles an hour, which is not a legal bike in Massachusetts anyway. And they just can't handle the impact. Yeah. Well, and you talked about one specific bike.

15:50Can you briefly tell us, because there's three different classes across the country, and which ones kids are supposed to be on if they're, you know, you don't think they should be on any of them. But talk about the speeds of all three. and something called jailbreaking, where they can actually get in and change the speeds. Yeah. So federally, there's three classes of e-bikes. Class one goes 20 miles an hour. You have to pedal. It gives you assist up to 20. Class two is the same thing, but it also has a throttle. So if you don't want to pedal, you can hit the throttle. It used to be that you had to get up to five miles an hour pedaling before the throttle would work.

16:24But nowadays, you can just sit on the bike and hit that throttle and go. Class three, which go up to 28 miles an hour, but the throttle will only get you to 20. You have to pedal that assistance to get up to 28. The big problem is a lot of the lower quality or Chinese made e-bikes coming in via different online sources, those bikes can be jailbroken. And they might say that they only go 20 miles an hour, but they can go on YouTube and look at a video on how to jailbreak that and make it go faster or take off the speed limit. Even on some of the nicer quality bikes, there's companies in Europe and places that sell devices you can attach to your bike to remove the restriction, which voids all warranties on them, but people don't seem to care.

17:05They just want to go faster. We're reading that some parents are saying, what's the difference? They can go fast on a bike, go fast down a hill. There's terrible traffic accidents with people on bikes. That's true. How would you respond to that? Well, pedaling a bike, a child is not going to go 20 miles an hour or 28 miles an hour or faster. They might go 12 or 15 miles an hour, but that'd be a kid in very good shape. But even then, bikes are much lighter. A bicycle typically weighs between 20 and 30 pounds. These e-bikes are taking brakes designed for bicycles, low-quality ones, and putting them on these big, heavy electric bikes.

17:41There's more to that conversation about e-bike safety, and you can listen to it all right now at hereandnow.org, where you will also find a conversation with two union leaders in the video game industry. Workers just notched a big win at Blizzard Entertainment, the studio behind hit franchises like World of Warcraft, Diablo, and Overwatch. They just ratified a first-ever union contract covering all 1900 company workers. This is an industry that had nearly no union representation just a few years ago. Workers Maureen Fatima and Paul Cox talked about the new contract with Scott Tong, Marine told him it's a big deal.

18:19I feel that it is a historic union contract, not just for the game industry, but for the tech industry at large. Some of the things that are absolutely precedent setting are, for example, universal recall rights, the ability to be recalled into any open position that you're qualified for across Blizzard bargaining units. If you get laid off, there's a provision in this contract that says if there is an opening that you're qualified for, you can compete for that position? Oh, more than that. If there's an opening that you are qualified for, you will be able to just get that job. You do not need to go through any interview process.

18:57I see. And Paul, let me ask you about Work Life Path Company. This is a company that had almost no unions five years ago. It had faced accusations of a toxic work culture. For you, what needs to change about the everyday culture of the place? So I've worked at Blizzard for 19 years in both Europe and the US. And I think that Blizzard's actually done a very good job in trying to improve the work-life culture. And one of the many things that we got in the contract was solidifying the things that we really liked about Blizzard. We have the holidays. We have a vacation from Christmas Day to New Year's Day.

19:36Very common in Europe, not so common here. We also have very flexible schedules. Something as simple as, hey, I've got a plumber coming over today. I'd like to work from home today, that sort of thing. Let me ask about jobs more broadly in this business, Paul. Tens of thousands of game workers have lost their jobs since 2022. The parent of Blizzard, Microsoft is promising more layoffs. Some could come in the games division. This is a business that makes more money than Hollywood and the recording industry combined. Why is the job market contracting? um if i were to guess it's because the industry got over invested in during covid where it absolutely boomed because everybody was at home and couldn't leave their houses it's also kind of similar to hollywood nobody wants to put out the game that does well they want to put out the game that does the best folks are basically like well am i going to be making 10, 30 times my budget on this?

20:35If not, I don't even want to invest in it. And I think that that's one of the problems that we're seeing right now. We're seeing indie studios, for example, who are doing very well. And yeah, they're not making a billion dollars on a game, but actually they're making 50, 100 times their budget because their budget was very low and they had passion. They were able to put in what they needed to get that game out. Listen to that whole conversation about unionization in the video game industry, and so much more at hereandnow.org.

21:08But that'll do it for the podcast today. Here and Now Anytime comes from NPR and WBOR Boston. Today's stories were produced by Joel Ryan, James Mastromarino, and me, Chris Bentley. Our editors were Todd Munt, Micaela Rodriguez, and Michael Scotto. Technical direction from Caleb Green and Matt Reed. Our theme music is by Mike Moschetto and Max Liebman, as well as me. Our digital producers are Alison Hagen and Grace Griffin. And here and now's executive producer is Alan Price. Thanks for listening. We'll be back with you tomorrow.