{"text":[[{"start":5.2,"text":"When Ned Ludd and his band of weavers took hammers to the new looms in early 19th-century England, mill owners retaliated by shooting them. I wonder how the tech titans will respond if voters switch from holding placards outside data centres to smashing them up. This seems likely, given the gulf between the pioneers’ excitement about what AI might offer, and the scams, slop and job losses which citizens currently see."}],[{"start":32.35,"text":"While some tech bosses talk inspirationally about scientific breakthroughs, other plutocrats breach copyright, stand by while their social media platforms harm mental health, demand more energy to power their ambitions with scant concern about climate change — and then boast with apparent glee about the eradication of white-collar jobs. "}],[{"start":52.6,"text":"Yes, the west must try to win the AI race with China on national security. But if that contest simply enriches a few, at the expense of the many, voters could conclude that we might as well embrace communism. "}],[{"start":66,"text":"Fear and distrust of AI has grown substantially since the release of ChatGPT over three years ago. Half of Americans now believe that AI will reduce our ability to form meaningful relationships, and almost three quarters feel AI development is moving too fast, according to YouGov. In the UK, a new King’s College London study finds only 24 per cent of citizens think AI is positive for humanity and a third of university students believe that AI will eliminate jobs fast enough to provoke civil unrest. The students who booed ex-Google boss Eric Schmidt’s graduation address this week may be first to mount the barricades. "}],[{"start":106.75,"text":"It might help if we asked what society actually wants from this transformative technology. What problems could it solve for the benefit of all? Could we employ AI to finally crack the problem of nuclear fusion, to bring abundant clean power in a world being rocked by climate change? Or to find a cure for Alzheimer’s, a source of so much human misery? Yet companies devote more time launching consumer products than talking about such quests. "}],[{"start":135.45,"text":"There is also an opportunity to make public services more efficient. AI can relieve social workers, doctors and nurses of form filling, boost personalised learning and speed up information delivery."}],[{"start":147.39999999999998,"text":"Despite the chance to use AI for good, the reality is that things are spinning out of control. The tech bosses feel compelled to go faster, despite being fully aware of the risks. Anthropic’s claim that its latest Claude Mythos model was too dangerous to release to the public has one upside: it has shown how close the world could be to an AI accident. The White House’s proposed new executive order on AI safety may be only voluntary, but it shows the libertarians are waking up. "}],[{"start":177.24999999999997,"text":"The willingness of Anthropic to restrict Mythos, and its creation of Project Glasswing — which brings together several tech giants to work on cyber security — offers hope that guardrails can be built."}],[{"start":190.49999999999997,"text":"Britain’s AI Security Institute, set up three years ago under Rishi Sunak, is now a world-leading AI safety body and has spawned others. "}],[{"start":199.84999999999997,"text":"But commercial pressures are outrunning safety concerns. Much of the policy debate still has a naive kind of deus ex machina flavour to it. The winner-takes-all philosophy of Silicon Valley makes it highly unlikely that competitors will co-ordinate on safety if it puts profit at risk. Ideas for mitigating the labour market effects of AI — more apprenticeships, taxes on capital rather than labour, a universal basic income — feel inadequate. "}],[{"start":226.39999999999998,"text":"Are fears of job losses a Luddite fallacy, or will AI permanently displace humans? It’s not yet clear. But public fears of job losses are rational: Standard Chartered announced the end of 8,000 roles only this week and described it as replacement of “lower-value human capital” with AI. One obvious response would be to create new trade unions. After all, the Luddites were against not just the new technologies but the lower pay and poor working conditions which came along with them. Strong arguments can be made for workers to collaborate with AI, not be replaced by it: this has already happened in radiology. If politicians want to give at least some reassurance to citizens they should ban civilian autonomous vehicles. There are simply too many people who rely for their income on driving jobs, and the benefits of replacing them are unconvincing."}],[{"start":278.29999999999995,"text":"Nerdy geniuses who wax lyrical about science fiction don’t help, by conjuring dystopias whose common theme is the loss of human control. As the most visible incarnation, it’s hardly surprising that data centres have become the focus of public discontent. "}],[{"start":295.29999999999995,"text":"Generative AI is pushing up energy consumption: in Ireland, data centres are now using 22 per cent of all electricity, up from 5 per cent in 2015. AI may eventually be able to mitigate its own environmental impact, but not yet. The US-based Center for Biological Diversity estimates that if current trends continue, data centres may account for nearly half of all US emissions from the power sector which national climate targets allow. Since August, at least 12 US states have debated temporary bans on data centre build-out."}],[{"start":330.54999999999995,"text":"Voters won’t wait forever for someone to slow things down: they may take things into their own hands. The 21st-century Luddites will be the middle classes and the working classes. And it won’t be possible to shoot them all. "}],[{"start":346.1499999999999,"text":""}]],"url":"https://audio.ftcn.net.cn/album/a_1779521303_7479.mp3"}