Across the City, a quiet shift is under way in how financial services professionals relate to artificial intelligence. Rather than waiting for employers to hand down training programmes — or bracing for redundancy — London's finance workforce is folding AI into daily work on its own initiative, and treating fluency with the tools as a career asset.

That is the picture painted by new Bloomberg research covering 500 professionals working in the capital's financial sector, alongside commentary from Amanda Stent, Head of AI Strategy & Research in Bloomberg's CTO Office.

Adoption is no longer the question

The headline finding is how thoroughly AI has already embedded itself in day-to-day finance work: 88% of respondents use AI tools at least weekly, and often daily. Strikingly, not one person surveyed said they had skipped learning the tools altogether, and nobody dismissed them as offering no upside.

Career incentives appear to be doing much of the driving. Some 85% of the workforce believe AI capability will soon count for more than conventional skills when promotions or new roles are on the line — a dynamic that makes keeping up feel less like a choice and more like table stakes. The pressure extends to peers, too: 69% of those polled take the view that colleagues who refuse to build these skills don't belong at their firm.

Learning by asking the machine

How are these skills being acquired? Mostly outside the classroom. Professionals report drawing on more than three learning methods on average, spread across both working hours and their own time — and the single most popular route is the most self-referential one: using AI tools to learn about AI, cited by 53% of the workforce.

Independent experimentation follows close behind at 50%, with 44% turning to videos and online guides and 42% picking things up from colleagues. Formal training laid on by employers comes last at 40% — underlining just how self-directed the City's upskilling push has become.

Stent argues this is precisely the right instinct. In her view, getting good with AI doesn't demand a computer science doctorate — it takes curiosity paired with domain knowledge. What stands out in the data, she suggests, is the initiative professionals are showing: learning by doing, tinkering, and putting questions directly to the tools themselves — exactly the disposition an AI-driven industry rewards.

More productive, not less busy

The research surfaces a productivity paradox. Fully 90% of professionals say AI has made them more productive — yet nearly three in four report their workload hasn't actually fallen. The hours AI frees up aren't disappearing; they're being redeployed.

Among those who have reclaimed time, 51% are channelling it into higher-value work — strategy, decision-making and client relationships — while 48% have cut back the time spent on routine administration. Another 36% say they're collaborating more with colleagues. Only 32% report simply finishing the day earlier.

Pressed on whether rising output without shorter hours just swaps manual grind for mental fatigue, Stent takes a measured view: the early gains are real, she says, but how they'll be shared out over the long run remains an open question. What she finds reassuring is where the saved time is going — toward work that leans harder on human judgement and expertise, not away from it.

Crime worries eclipse job fears

For all the public hand-wringing about automation displacing workers, London's finance professionals rank job losses near the bottom of their worry list. Asked to pick their top three AI concerns, only 30% flagged the erosion of entry-level and graduate roles — while 45% pointed instead to AI-enabled cyberattacks and financial crime.

That relative calm reflects a shared understanding of where AI's limits sit. The technology strips friction out of manual research, but it can't shoulder accountability. As Stent frames it, when machines handle more of the retrieval and synthesis, human effort migrates up the chain — from performing the research to serving as the final arbiter of what it means. Any government blueprint for AI adoption in financial services, she adds, needs to plan around that shift.

A model can flag a pattern. It takes an experienced trader or analyst to know whether that pattern is signal or noise. — Amanda Stent, Head of AI Strategy & Research, Bloomberg

The workforce's enthusiasm, in other words, comes with a firm boundary: critical financial decisions stay under human oversight, with strong consensus that autonomous decision-making has no place in core functions. Stent's closing observation captures the mood — in an era of abundant information, judgement and leadership are the scarce commodities, and the City's professionals appear to grasp that and are positioning themselves accordingly.