BenevolentAI's Founder Ken Mulvany Talks AI Risk

SPEAKER_1 (0:02)

Bloomberg Audio Studios, Podcasts, Radio, News.

Caroline Hepker (0:07)

Now I want to talk about UK startups, funding and AI. This at the start of Cambridge Tech Week. So the new UK Chancellor John Healy wants to double the number of unicorns. The UK has got three of the world's top ten universities and lots of spin outs. But still, what experts say is a persistent technical literacy gap.

So with lots of mature spin outs increasingly leaving the UK altogether. Joining us now is BenevolentAI founder, Kenneth Mulvany, who is the keynote speaker at Cambridge Tech Week, that starts today. Ken, good morning. Thank you for speaking to us on Bloomberg Radio.

Kenneth Mulvany (0:47)

Good morning.

Caroline Hepker (0:49)

So I just want to start by talking about AI, because the big news story today is about Anthropic and OpenAI promising these measures to sort of rein in cutting edge AI models, have third party sort of oversight within the businesses. I wonder as an AI tech entrepreneur yourself, what you make of that?

Kenneth Mulvany (1:10)

Well, I mean, AI has been around for a very, very long time, but the power that we're seeing with the models right now are opposing, at least from Benevolent's perspective, are opposing all sorts of different sort of threats that we didn't have maybe five, ten years ago. And some of these are around, well, use or misuse of these models in biology. They're becoming more powerful, more accessible. And so it creates enormous opportunities for us all, but it also creates enormous changes on the risk landscape. So I've been watching it very closely.

I don't feel the existential threat that we're seeing so much in the media. Well, maybe I do, but there's more homegrown things that I'm worrying about, which is the misuse of this technology rather than this technology becoming autonomous and having a grudge against humans.

Charles Capel (2:27)

Yeah, I mean, I guess that's why some sort of safeguards could be useful. I mean, one of the scenarios that people talk about when we talk about the potential for AI to harm, or in the past few days, we've been talking about killing humans, is around this idea of AI creating a virus, or messing somehow with biology. Now, BenevolentAI is in the biopharma space.

Do you think that your researches could do with a little bit more breathing space or a little bit more regulation to try and slow this down, and crucially so that they understand exactly what the AI is doing?

Kenneth Mulvany (3:13)

Really good question.

It doesn't take a lot, right? It doesn't necessarily take creating a new virus, right? You can change attributes of a virus, make it more contagious, or have a longer latency period between when you are sick and infectious to when you start showing symptoms. And those are all within the realm of today's technology. And it's probably now maybe a year ago, they began to shut some of these things down and look at this much more closely.

I know that Anthropic published something not too long ago where it elicited the number of things that it has shut down, where it felt that it was beginning to get dangerous. And Benevolent's working increasingly with governments on this. You know, the type of things that we did during COVID, you know, that created a treatment for COVID. This was parasitical for very sick patients that were hospitalized. It took about 48 hours for it to do that. I mean, it took a very long time for it to get through its clinical trials and get, well, actually to take that one, took about a year for it to get through the trials and ultimately approved. And that sort of capability we're now bringing to bear for other pandemics, but also kind of the notion that there could be bad actors that are out there. And I think that government is looking increasingly hard at these things and ourselves and some of the companies that you've mentioned and others are part of that consortium looking at this very closely.

Caroline Hepker (5:12)

Yeah, I think that's really interesting and we appreciate you giving us some answers on this because actually Dario Amadei's essay over the weekend sort of raised more questions, hasn't it, for so many people because this whole idea of trying to have coordination between democratic governments and authoritarian governments, between having, you know, people actually inside with kind of employee level access so that these models can be looked at. And I suppose how that intersects with your expertise in life sciences, I think is just very, very interesting.

But I suppose on a bigger point also for the UK, what's your view on actually how, you know, and this is more relevant to Cambridge Tech Week, I suppose, what is your view on the landscape for spin outs from UK universities, which I wanted to kind of get your take on. You know, the government now wants to try to double the number of unicorns. They haven't actually put a date on that. They've just said they want to double it. So we don't have a timeline. What do you make of the tech eco and spin out ecosystem now in Britain?