{"text":[[{"start":6.45,"text":"A few months ago a New York financier told me he had just experienced a “first”: his 2025 summer interns “were the first true AI natives I have seen”. This meant they had grown up not only among digital tech, but AI too."}],[{"start":21.25,"text":"So how did it go? He winced. While those wannabe masters of the universe initially seemed wildly impressive, when senior financiers later probed their ideas they found them alarmingly shallow."}],[{"start":33.6,"text":"Consequently this person’s company made fewer return offers and is now focusing less on graduates in science, technology, engineering and mathematics — and more humanities students instead. "}],[{"start":44.55,"text":"“We want critical thinking, not just AI,” he explains. Human brainpower is needed to handle the silicon variant."}],[{"start":51.599999999999994,"text":"Now, this is just one tiny anecdote in a tsunami of emerging AI-linked stories and I daresay there are counter-examples. But it illustrates a bigger point: as AI fever sweeps finance, it is neither delivering the profit nirvana predicted by tech evangelists nor hastening the doom the Cassandras have warned about. Or at least not yet."}],[{"start":72.35,"text":"Yes, techies are wildly excited about putative productivity increases, with entities ranging from JPMorgan to Visa now presenting themselves as AI tech groups, not stodgy money handlers. Indeed, a recent survey from Nvidia claims that 89 per cent of finance executives say that AI is boosting revenues, while 73 per cent think it’s “crucial to their future success”."}],[{"start":97.25,"text":"Unsurprisingly, that is alarming regulators too. One reason is that the eye-popping volumes of (mostly) private credit backing for AI data centres is likely to produce defaults, as the Financial Stability Board warned this week. "}],[{"start":111.5,"text":"But the other risk is that the technology itself damages financial stability. The immediate threat on financiers’ minds, as the IMF notes, is a systemic crisis arising from a cyber attack. But as the FSB observes, there are other dangers: computer “herding” around identical trades; concentrated reliance on a few cloud providers outside regulatory oversight; and hallucinations from models."}],[{"start":135.4,"text":"Regulators are responding. Scott Bessent, US Treasury secretary, and Jay Powell, Federal Reserve chair, recently met top financiers to discuss the threats posed by Anthropic’s new Mythos model. And Michelle Bowman, Fed vice-chair, has called for more “flexibility” in how regulators evaluate banks’ risks models. "}],[{"start":155.70000000000002,"text":"This week the UK’s Financial Conduct Authority took a novel step: it is offering free computing power and data to fintechs that work inside an FCA “supercharged sandbox”, after the success of a pilot scheme with Nvidia last year. “The idea is to have an environment where [innovators] come in and test and bring their solutions to life in a safe and responsible way,” explains Jessica Rusu of the FCA."}],[{"start":182.55,"text":"But as the AI rhetoric intensifies, AI reality is proving to be more complex. Consider, as one example, a new survey from the Judge Business School in Cambridge of 628 finance and AI companies and regulators from around the world. (Disclaimer: I work at the University of Cambridge, but was not involved in this research.)"}],[{"start":203.25,"text":"This found that 81 per cent of private-sector finance groups are now using AI, and half are embracing agentic AI. Most of these firms use external models and the top provider, by far, is OpenAI. "}],[{"start":216.2,"text":"However, the survey also suggests AI is overwhelmingly being used for back-office functions, with limited deployment elsewhere. And, unlike the Nvidia poll, the Cambridge survey found that 76 per cent of large financial groups — and 55 per cent of all industry respondents — struggle to measure the value of AI. "}],[{"start":237.6,"text":"Indeed, only 40 per cent reported any profit boost from AI and 43 per cent saw no change at all. More notable still, only a quarter expect sector job losses, while 58 per cent actually expect AI to spark more hiring or reskilling at their own group."}],[{"start":256.55,"text":"AI evangelists might dismiss this as proof that some financiers are Luddites. After all, the AI revolution is still in its infancy. Meanwhile, many regulators seem even more Luddite: their level of AI adoption (and thus knowledge) is half that of the private sector groups they are overseeing. This is worrying. "}],[{"start":274.8,"text":"But the point about the gap between rhetoric and reality in AI deployment is echoed in numerous other surveys from McKinsey, the Institute of International Finance, EY and others. A survey from Hyland, for example, says that only 45 per cent of businesses say AI is delivering the results they hoped for. "}],[{"start":294.35,"text":"This may be merely a time lag effect. But I suspect it also reflects something else: intelligent machines do not automatically deploy themselves either for good or bad. Human strategy is crucial."}],[{"start":306.35,"text":"So the big question now — for financiers, regulators and educators alike — is to how to create skilled AI natives who can also use the critical thinking needed to spot both the opportunities and (very real) risks in AI. Those companies which find them will be the real winners. All eyes, then, will be on the 2026 intern pool."}],[{"start":334.00000000000006,"text":""}]],"url":"https://audio.ftcn.net.cn/album/a_1778847932_1890.mp3"}