The Future with Hannah Fry on BBC Two: Just as humans stuck gorillas in zoos, could AI do the same to us?

Hannah Fry was hopeless on Celebrity Traitors, where all of her scientific discipline and boffin’s aura evaporated the moment she was made a Traitor and had to go around back-stabbing her fellow A-listers. But she is on firmer ground with the agreeably informative yet informal second series of The Future with Hannah Fry (BBC Two, 7.30pm), where she ditches her Traitor’s cape and once again tackles the big science issues of the day.

These subjects include the suddenly quite-terrifying issue of AI. In the previous season, she investigated the chilling subject of chatbots, and had a very weird encounter with a man who had constructed an AI bot based on a dead parent, so that he could continue to converse with them from beyond the grave.

That was Black Mirror with rocket jets attached – and it had the usually level-headed Fry in tears of shock and mild horror. That said, talking to weird personalities who only vaguely approximate real human beings was the perfect preparation for Celebrity Traitors and, to that extent, the experience presumably stood to her.

The going is comparatively cheerier as this latest run of episodes begins with a deep dive into the subject of machine learning. Fry begins her adventure by gazing at some gorillas in a zoo in London. Humans and gorillas share a distant ancestor – yet humans have gone on to enthusiastically bung their fellow apes into enclosures and destroy their natural habitats.

The obvious worry, she notes, is that AI could do the same with us. Gorillas have only a vague awareness that the world they live in is designed by and for humans. Might AI wreak similar havoc – with our puny-ish brains unable to process what’s happening?

[ AI: are we building a partner, or a Frankenstein’s monster?Opens in new window ]

Fry hasn’t quite made up her mind, and reminds the viewer that AI chatbots such as ChatGPT and Elon Musk’s beloved Grok aren’t thinking machines as we might imagine them, but vast spreadsheets. She delves into the deeply unsettling possibility that, to become fully intelligent, as we think of the word, machines would have to exist fully in three dimensions – ie, acquire some kind of a robot-type “body”.

“ChatGPT can guess what gravity is – based on people’s descriptions of it. It’s a reflection of a reflection. Whereas, if you actually experience it, you get it right from the source,” explains Sergey Levine, an associate professor at UC Berkeley.

Concerns over the possible impact of AI on humanity have been given short shrift by the tech industry – and that may come back to haunt us, says UC Berkeley’s Prof Stuart Russell. “If you make machines that are more intelligent than we are, it’s not going to be easy to retain power over them forever,” he says. “We could just decide not to do it. But the economic incentives are too great. The amount being invested right now, specifically to create superintelligent AI, is in the ballpark of what the entire world spends on basic science research.”

[ Is artificial intelligence the greatest threat to humankind?Opens in new window ]

Fry also explores the issue of racial bias in AI. “We’ve seen problems with facial recognition software. It makes more mistakes with people who have darkened skin colour,” says Melanie Mitchell, the James B Alley jnr professor at the Santa Fe Institute in California.

Fry reminds us that the human brain is many times more sophisticated than the average chatbot – and that it is by no means a given that we will end up locked away, being gawped at by the machines. “While the gorilla problem is a poignant warning for the future – we should not be distracted from today’s risks, like racial bias and fake news.”

It is hardly an optimistic conclusion. But Fry nonetheless makes an important point about the threat posed by AI not being some dystopian nightmare but one rooted in the same old prejudices that have afflicted humanity for centuries. Just like the poor old gorilla in the zoo, we can’t escape from who and what we really are.