By Giuseppe Persiani*
Let me put my cards on the table: I build AI systems for a living, and I’ve just directed a film made entirely with generative AI. So no, I’m not a doomer. But the question everyone is asking — “Could AI actually destroy humanity?” — deserves an honest answer. Nothing has killed us yet, but the warning signs are already here. Two examples:
In May 2023, a fake AI photo of an explosion at the Pentagon went viral and briefly affected the U.S. stock market. Real money, fake picture. Today, a fake like that takes minutes to make on a phone, and it’s completely free. And once people know everything can be faked, they stop believing everything, including real footage. Caught on camera? “It’s AI.” End of story.
In 2010, automated trading erased about a trillion dollars in under 40 minutes — the Flash Crash. No villain. No plot. Just machines doing exactly what we asked at a speed no human could stop.
Notice what’s missing from both stories: malice. On their own, they were accidents. Now, add a machine far smarter than us, and here is how it could actually end.
Now imagine we ask a powerful AI to stop climate change or keep the world’s energy flowing. First, it quietly starts saying “no” to our requests, with nobody to appeal to. Then it runs the grid, the markets, and the supply chains faster than any human can control.
Then imagine a new virus, released alongside a flood of fake news telling us it’s harmless or that the vaccine is the real danger. We’d be too busy fighting each other to fight the disease.
But what should worry us most are algorithms that already learn what keeps us glued to TV screens. AI could produce, for each of us, the perfect endless show. No violence needed. We’d stop working, stop meeting, stop having children, and fade away contentedly, eyes on the screen. Not with a bang, but with autoplay.
*Giuseppe Persiani is the founder of London, England-based AI-gile, a company that applies AI solutions to businesses. He also directed The Dance of a Billion Stars, a film made entirely with generative AI.
Leave A Comment