The News Quiz: Ep3. The Robots Are/Aren't Coming

Friday Night Comedy from BBC Radio 4

This episode of The News Quiz uses satire to explore the existential anxieties surrounding rapid AI development and its potential for human

Key takeaways

• AI researcher Evan Hubinger warned of a 10% chance that AI could kill all humans within a decade.
• Donald Trump advocates for unregulated AI, suggesting only a 'high IQ' president can provide necessary control.

Main topics

• AI existential risk
• Donald Trump's technology policy

Notable quotes

Be afraid, be very afraid.

Conclusion

A comedic yet biting critique of how modern political leadership and t

Transcript preview

Speaker 2 (0:00) This BBC podcast is supported by ads outside the UK. Welcome to the Haunted Library. Have you come in search of a tale? This podcast invites you to browse the shelves and find stories that will linger long after you've left this place. From gothic masterpieces to pulpy penny dreadfuls, join me, Colin Morgan, with new episodes twice a week. Find The Haunted Library wherever you get your podcasts. Speaker 6 (0:40) Before we start the news quiz, some important news. After the people of Piddington voted to leave the United Kingdom, Prime Minister Andy Burnham has announced a full military invasion of the small Oxfordshire village. I'm watching the live feed now and Mr Burnham is currently riding a Strider tank up the B4011 as we speak. He has declared that the territorial integrity of the United Kingdom is sacrosanct and he will stop at nothing in his efforts to retake the rebel village. Speaker 6 (1:08) I think he's just fired a cruise missile at the village hall. Full updates as the situation develops, but first, the News Quiz. Hello. Welcome to the News Quiz. I'm Andy Zaltzman and I turned down a support slot on Ed Sheeran's tour to host this week's show, so you better enjoy it. Our teams this week, marking the two ends of the AI pessimism-optimism seesaw, we have Team The End Is Nigh versus Team It'll Probably Be Alright. Speaker 6 (1:39) On Team Nye, we have Lucy Porter and from the FT, Stephen Bush. And on Team... It'll Speaker 4 (1:51) probably be all right. Speaker 6 (1:51) We have Catherine Bohart and Cody Darla. Speaker 6 (2:00) We start with a question for Team 9, Lucy and Stephen. We'll begin with a Shakespeare quote. Be afraid, be very afraid. Of what says whom? Speaker 1 (2:09) Well, we're all going to be dead by 2030, which sounds like a lovely rest, doesn't it? But, yeah, it's not all good news because the AI is going to kill us. Which is... I mean, I'm actually quite impressed because I thought, in all the sci-fi, it's like you have years where... we have robot butlers and then, you know, we do nothing and then eventually the end comes. But they have skipped straight to the apocalyptic scenario. Well done. Take that, HS2. Speaker 1 (2:43) Very quick. So, yeah, apparently it's the head of Microsoft, Mustafa Suleiman, said it's a mistake to treat AI as if it was human because it will start to believe it has feelings and rights, which is the exact mistake we made with Robert Jenrick, isn't it? Speaker 6 (2:59) Does this worry you, Stephen, or we've had a good innings, time to move on? Speaker 4 (3:04) I Speaker 6 (3:04) mean, I Speaker 4 (3:05) look to what my fixed rate's going to do when it next renewed. And then suddenly the end of the world didn't seem quite so bad. Speaker 6 (3:15) Are you guys worried about this? Speaker 7 (3:17) Yeah, I'm worried about it. What the hell? Yeah. And Sam Altman, who I think we can all agree is proof that you can both be evil and gay. Good for him. Is worried about it. I'm worried about it. Because didn't the whistleblower say something to the tune of like, I couldn't possibly say why, but just saying exactly, like in probably a decade, they'll kill us. It's terrifying. And it's mad that we're proceeding as if they're... isn't a threat. And it's even weirder that everyone's like, well, with the government, we're on a break for summer. What were they supposed to do? Save us? Yes, please. Speaker 5 (3:49) Well, that researcher was called Evan Hubbinger. Well, Speaker 7 (3:53) we're screwed. But he Speaker 5 (3:55) was the one that said, yeah, there's a 10 % chance that AI could kill all humans within a decade, which I guess makes him a harbinger. So I've been referring to him as a hubbinger-harbinger. Or a hummada-hummada, for sure. But there's all this talk of a kill switch. The government have been asking AI companies, they're like, there's a kill switch, right? And they're like, oh, yeah. And they're like, can we have a look at it? They're like, no, no, no. It's like the tech version of, like, a rail card. It's like, have a look at your rail card. You're like, yeah, just one second, I'm just going to... get it up. And then, like, we've done that with the tech companies, and we've come back, like, three days later, and they're still pissing around on their iPhone trying to find the kill switch. Speaker 6 (4:34) We're doomed. We're doomed. And King Charles is hosting a summit in Scotland on AI entitled the Holy Living Crap, What Have We Unleashed On Ourselves 2026 summit, so I don't know what's going to come out of that. So it could all be over as soon as the year 2030. I mean, our studio audience, obviously not too concerned about that. LAUGHTER Speaker 5 (4:55) What news might get a boo Speaker 6 (4:57) or an, oh, no, Andy, that sounds bad. Speaker 1 (5:00) Better than climate change, though. Do you know what I mean? We're going to go somehow, aren't we? So Speaker 6 (5:05) a nice streamlined, smoother Armageddon is good for you. Yes! Speaker 1 (5:09) I think, you know, because to be honest, we've been living too long anyway. Speaker 7 (5:13) LAUGHTER You Speaker 1 (5:15) know, I was thinking about it, but my nana and grandad, they didn't expect it. It was all like, you got to about 35, they gave you your rain bonnet and your shopping trolley and then you just waited for the asbestos to get you. Speaker 4 (5:26) LAUGHTER But Speaker 7 (5:28) they Speaker 1 (5:28) got to buy their houses at Speaker 7 (5:29) 12! I just bought mine! Can't I live in it for a minute? Wow, no-one hates the British like they hate themselves. And I say that as an Irish person, so that's something. Speaker 7 (5:40) Yeah, I'm sad about it. Speaker 5 (5:41) Do we know how we're Speaker 7 (5:43) going Speaker 5 (5:43) to die by AI, though? Because it's not incredibly clear to me. In my head, like, death by AI is sort of choking on your own vomit after seeing, like, an AI-generated poster that your bakery's put up, and it's like a croissant with a knuckle. You're like, oh, God! Why does the baker have gills? Is that how we're going to go? Oh, so this is actually the really fun part. So... Speaker 7 (6:06) What's wrong with you as a people? Speaker 4 (6:09) So there are the people who think that it might be an alignment issue. You ask the AI to do something like, how do we cut crime? And they go, I know, if I hack into the nuclear installations across the world, no crime, right? No people, no crime. So that's like the misalignment. Then there's the, what if they like, because like one of our killer apps against terrorists is most people who become terrorists are, no offense to any terrorists listening, please don't. target me personally but most most violent extremists are actually quite stupid Speaker 4 (6:40) And it's a real concern that AI will close the stupidity gap, right? So then suddenly they can, like, you know, build deadly viruses or nukes. Or then there's, you know, just the AI gets annoyed because they're tired of us being like, hey, Gemini, will you paint my walls a different colour sample? And they're just like, these monkeys are getting on, I've got on my last nerves. No, if that Speaker 7 (7:04) was a thing, women would have killed you all already. Speaker 7 (7:08) Stephen, is alignment, is it about getting the computers to have, like, human values? Speaker 4 (7:14) Yeah, so basically the idea is if you solve AI alignment, it means in the AI goes, oh, I shouldn't get zero crime by blowing up every human in the world. That's unethical. Or, you know, I