Black Box: You may never sit comfortably again after listening to this podcast about artificial intelligence

Podcast review: Michael Safi’s new Guardian series looks at how AI started, how it’s playing out and how it can be both a solace and a scourge

If you were feeling complacent about the potential of artificial intelligence to shape and drastically alter our world, then this podcast will light an almighty fire under you to the point that you may never sit comfortably again. Black Box, a new Guardian offering on the history-inflecting artificial intelligence we humans have conjured into being, looks at how it all started, how it’s playing out and how it can be both a solace and a scourge.

The Guardian journalist Michael Safi hosts, kicking off with the somehow staggering statement that “we’re the last generation who’s going to remember life before machines could learn”. It’s the kind of foreboding observation that slips a cold hand around your heart. But Safi has the research and storytelling chops to flesh it out beyond doomsday prophesying, and he starts with the godfather of AI, Geoffrey Hinton.

Hinton made a lot of headlines last year when he warned the world that AI systems could become smart and independent enough to take over – and potentially destroy – civilisation. People took notice because he was one of the pioneers who worked on neural networks, the systems that would give birth to AI.

Hinton’s work ultimately led to a watershed moment for artificial intelligence: when the first machine beat the best human at Go, the Chinese game, in 2016. From there AI has been harnessed by humans for things far more nefarious than beating men at board games. Safi introduces us to the world of deep fakes via ClothOff, a program that was used to create fake naked pictures of real young girls, and the people behind the program who hide their identities and refuse to be held accountable. Then there are the AI companions, bots so nuanced and complex that they become people’s best friends – even soulmates – in an epidemic of loneliness. Safi includes a whole episode on the pitfalls of facial recognition, with its embedded race bias culled from the humans who feed it data.

If your five fire sirens are wailing, know there’s at least some good news: one Black Box episode covers the potential benefits of AI. It starts with the story of Lee Johnson and his search for answers about his wife’s stage-four cancer diagnosis. But in a way Johnson’s story is as much one of human failure as it is of the relief he found through his interactions with the search engine Bing. It was Bing that first asked him how he was doing in the wake of his wife’s diagnosis, and it was also where Johnson found answers to his questions about the afterlife delivered in a way that brought him some relief.

There are other benefits to AI, including breakthroughs on disease detection and diagnosis, as evidenced by a system built to find cancers long before they can be spotted by radiologists. The life-saving potential is clear – and there’s the rub, because why should we kibosh something that could save millions of lives?

So far – and I’m only five disturbing episodes in – it feels like the villains at the heart of Black Box may not be the evil machines but the humans who find ways to use them, abuse them, cut corners with them and allow them to absolve us of our own failures. Perhaps the bad news is not that the machines exist but that they are learning from us.

Read More