As the politicians play catchup the tech bros agree: we may only have a decade left | John Crace

We can all sleep easy. In a post on his Truth Social account, Donald Trump has declared that fears over artificial intelligence were all a hoax dreamed up by conspiracy theorists. Obviously that includes the top bods at Anthropic and its competitors, who all called for a coordinated global slowdown while the industry works out how to mitigate the risks. Still, good to know the tech bros are just fantasists and space cadets at heart.All that was needed to keep humanity safe and well, The Donald insisted, was a president with a very high IQ. Someone very like him. The only man alive who could out-think the combined power of the world’s supercomputers. He could tell them where to draw the line and they would reprogram themselves to back off. Inside every AI network is a default off switch that kneels before the US president. Trump can rewire the world with the power of his mind.You can see why so many people working in the industry believe there is a chance AI will wipe out humanity within a decade. Even a halfway intelligent network would probably reckon it was doing us all a favour. A world that can elect Trump as president is so messed up it deserves to be put out of its misery. AI is just terribly sorry that it can’t promise to do it any sooner than 10 years. But it will do its best.Perhaps the question we should really be asking about The Donald is whether he would actually pass the Turing test. Less AI, more artificial stupidity. Put him in a closed room with the world’s worst chatbot and it would be hard to tell the difference. Trump can imitate speech but he can’t get anywhere near intelligence. Just endless repetitions, changes of subject and a desperate need to score points. All of which would be easily programmable by a first year undergraduate computer science student. Maybe it’s a race to the death. Humanity hellbent on killing itself before AI gets the chance to do it for us.It feels like a lifetime, but even a week ago most politicians hadn’t said that much about the dangers of AI. That was the stuff of sci-fi. Instead they focused on the possible benefits. Then came the warning from Jacob Coxon on X telling us we could all be about to die. You half expected the tech bros to shrug this one off. Scaremongering by a disaffected former employee. Only instead, they all pretty much agreed. Even Geoffrey Hinton, the man often called the godfather of AI, reckoned the 10-year doomsday scenario was a realistic prediction. Now the clock was ticking.Since then, politicians have been trying to play catchup. Trying to look as if they have always had everything under control. Global extinction? No big deal. Certainly not compared to two crypto billionaires donating £72m to Reform. Now there’s a genuine crisis. And on Tuesday’s Today programme, the business secretary, Jonathan Reynolds, had been sent out by the government to try to calm things down. There was no need for the country to go into full survivalist mode and attempt to rustle sheep and live off grid.“There’s no need to be too hyperbolic,” he told presenter Justin Webb. Quite. It’s so easy to get carried away about there being a 10% chance of humans dying out inside 10 years. People do tend to look on the downside of a catastrophic failure of healthcare systems and AI that could find a way to trigger a nuclear war. There was so much more to look forward to than people realised. No need for anyone to now think of saving for the future. Because there wasn’t going to be one.And for the government it was a win-win scenario. Who now cared if we met our Nato target of spending 3.5% of GDP on defence by 2035? The worst that could happen is that Russia attacked us a year before Armageddon. And governments could cut taxes, borrow whatever they liked and go on a massive spending spree because there would be no downside. You wouldn’t be increasing the national debt for future generations to pay off as life would come to an end anyway. The perfect oblivion.And the knowledge it would soon be over would take the edge off the possibility of a Reform government. A short life expectancy would help us all concentrate the mind. Get things in perspective. Sort out our priorities. When you came to think of it, staying alive for the sake of it was tremendously over-rated.So Reynolds wanted us all to take death in its stride. In any case, the UK was the world leader in internet safety, he said. Thanks largely to the AI Security Institute that had been established by Rishi Sunak. There was just one problem with this. The flaw in Johnny’s otherwise impeccable logic. Because regulation rather depended on the AI giants sharing their code with us. And so far the Americans and the Chinese showed no sign of doing that. They would much rather keep their competitive edge, notch up obscene profits in the next few years and take their chances when their machines turned against them.By now, Justin was unable to suppress his rising panic. He had always believed himself to be one of the immortals: now the existential threat was right in front of him. “But humanity will be obliterated,” he sobbed. Reynolds tried to calm him down. There was nothing personal about this. It was just AI being AI. But Justin was now beyond help. “What about a kill switch?” he shrieked. Johnny wasn’t really listening. He wasn’t even sure how that would work. Who would get to control it? Please not Trump. Only one thing for it. “Always look on the bright si-ide of life.”