Taking Stock: Can Big Tech rivals be trusted to develop safe AI?

Will Bain (0:00)

Hello and welcome to Business Daily from the BBC World Service. I'm Will Bain, where we're taking stock of a business week that feels a lot more like the start of a sci-fi film. AI in general is good.

Katrine Kielos (0:12)

I don't think it will be the end of civilization.

Jack Clark (0:15)

Talk to anything about anything and it will end civilization. Always has.

Will Bain (0:18)

There's gotta be some kind of truth to it.

Katrine Kielos (0:20)

It is being developed at a very alarming speed.

Will Bain (0:24)

Trust the general public to take complete civilization annihilation in their stride in Washington there.

Our guest this week on the programme, Rema Rahman. Rema is the White House and Politics Editor for Bloomberg Government in Washington DC. And Katrine Kielos, author and columnist for the Swedish financial broadcaster EFN, is joining us from London where she works through the week. Hello guys. Thanks for being on the programme. Hi.

Rema Rahman (0:50)

Hello. Thank you.

Will Bain (0:51)

Rema, did you have a more perhaps dramatic take than our very calm BBC listeners who were there in a parking Washington DC responding to everything that was coming out from the big AI firms this week?

Rema Rahman (1:04)

You know, I think there's a kind of general consensus sort of joking about it, but also thinking, well, you know, if that's the case, then kind of shrugging their shoulders at it and thinking, well, maybe I'll skip the gym today and have that extra cookie. Why not?

Will Bain (1:18)

Well, that's the thing here, Katrine, isn't it? And it's really well summed up in that sort of 15 seconds or so we played of people, where they're trying to brush it off. But they also are a bit like, maybe if it is really serious, well, exactly as Rema says, well, I can't do anything about it.

Katrine Kielos (1:33)

Yeah. And I think the headlines are a bit confusing. And you see, anthropic researcher believes more than 10% chance AI could, quote, kill all humans, end of quote. I mean, I think you do think how did they land at that number? You know, it's very specific. Why 10%?

Why not go for 10.4?

I think it's hard to conceptualize. But I mean, obviously from a European perspective, I mean, we like to regulate things here. And the European instinct is, you know, if the people building the technology are themselves telling us there is a non-zero risk of losing control of it completely, then perhaps governments shouldn't simply say, you know, well, the market will sort this out. So and that's probably reasonable, but I also think Europe needs to understand that AI is not a toaster. I mean, we're very good at product regulation, but this is not something that you can check at the factory gate. Oh, this is safe and just send it out. It's a different thing. And I think all of these unknowns make it hard for people to grasp these risks. And therefore they brush it off, as we heard in the parking lot.

Will Bain (2:39)

Well, we're going to get into all of that there. The risks, the potential for regulation, what everybody's saying about it on this edition of Taking Stock from Business Daily.

Yeah, welcome back to the program. As we say, it's been the story of the week, the increasing loud, sometimes pretty stark warnings from inside the AI industry, which I guess has been the unusual bit about it, about the dangers AI might already and in the future be posing towards us. It all started, if we take you right back to the weekend, with three of the biggest names in AI, Elon Musk with his SpaceX company and Grok Chatbot, of course, Sam Altman of ChatGBT, the owner of OpenAI, joining calls from the boss of Clawmaker, Anthropic Dario Amadei, to slow development of AI to allow regulators to catch up a bit. That was quickly followed by claims, as Katrine was just mentioning, from one of Anthropic's former staffers, Jacob Coxon, telling us here at the BBC at the weekend, researchers were genuinely frightened about the speed of its advancements and what it could mean for humanity. One of the co-founders of Anthropic, Jack Clark, spoke to our colleague Faisal Islam, and he said AI companies should be forced to create and use kill switches to keep rogue AI agents in check.

Jack Clark (3:53)

If we continue as we are, which is basically a broadly totally unregulated industry developing incredibly powerful technology that is getting much, much smarter all the time, likely soon smarter than people, then I think that we are rolling dice with immense risks. And the point is, we have to change the course of this industry. We have to get a policy regime in place that allows us to have greater control over the industry and a better understanding of the safety and security. And then you can reduce that chance of there being a major risk to society.