UBS’s historic headquarters at Bahnhofstrasse 45 in Zurich. The bank’s 2027 graduate intake will be the first to face the new requirement. Credit: Ank Kumar / CC BY-SA 4.0 via Wikimedia Commons (cropped) For two years the AI story in banking has been a headcount story. How many graduates a bank takes, and how many it quietly stops taking.
UBS has changed the question. It is no longer asking how many juniors it needs. It is asking what a junior has to be. The change covers anyone applying to start at the Swiss bank in 2027, as a graduate or an intern in Global Banking and Markets.
Those applicants will have to show that they can use AI to improve outcomes and efficiency. The Financial Times first reported the change, citing people familiar with the matter. Interviews will include questions about how candidates actually use the tools. TNW has not independently verified the account.
The bar is not whether an applicant has opened ChatGPT. It is whether they can point at a real task and explain what they handed to a model. Then say what came back that was better than before. That is a harder thing to rehearse than a competency answer, and it is meant to be.
What UBS is actually asking for The requirement sits alongside the usual entry criteria, including a strong degree, rather than replacing them. UBS told the Financial Times that AI skills complement academic and social abilities. The same standard will apply to other newly posted roles. What UBS has not said is whether the requirement will eventually reach the staff it already employs.
Successful candidates then go through an internal programme UBS calls the AI Fluency Pathway. It covers real banking use cases and responsible use. That detail matters more than the headline requirement. A bank that expected finished expertise on day one would not need to build a training track.
What UBS wants at the door is evidence of use and a willingness to learn. It is among the first large banks to write the requirement down. Santander wants advanced AI users on some of its trainee programmes, according to the Financial Times. The practice is spreading rather than starting.
Every story before this one pointed the other way This is the first move in the run that is not about shrinking the intake. In August, EY stretched its internship to a year before offering a job. That bought the firm time to work out what a graduate is worth. A Swiss study in June found fewer job ads aimed at career starters.
That is the same labour market UBS recruits from. Harvard researchers found AI-native startups hiring fewer juniors and more senior specialists. Each of those is a story about supply. UBS has made a story about specification.
The graduate job still exists at the bank, and the price of it went up. The numbers behind the nerves Banks have spent the past year automating the work that used to fill a first year. Financial analysis, research, and the assembly of client presentations. UBS is also testing analyst avatars for client presentations, the Financial Times reported.
That is the same job viewed from the other end. Morgan Stanley analysts expect more than 200,000 banking jobs in Europe to disappear within five years. Separately, in May, the bank doubled its forecast for how deeply AI would cut European banking headcount. That revised figure ran to as much as a fifth of all jobs.
Jamie Dimon has already said AI eliminated 30 to 40 percent of jobs in some JPMorgan divisions. Against that, a hiring requirement reads almost like reassurance. UBS is still recruiting graduates. It has simply decided which ones.
For a candidate staring at those forecasts, being told what to learn is better news than being told there is no intake, even if the requirement itself arrives as one more thing to fund out of their own pocket. The part nobody has solved There is a catch inside the policy, and a senior banker named it first. Conor Hillery, who runs JPMorgan’s business in Europe, warned in December that juniors cannot afford to lose the fundamentals. He was pointing at the obvious risk.
An analyst who reaches for a model before understanding the problem may never learn to understand the problem. The research supports him. A July study described the effect as never-skilling. Juniors who lean on AI from the start often fail to develop the debugging instinct that comes from getting things wrong slowly.
Screening for AI proficiency at the interview stage does nothing about that. It may even select for it. UBS appears to know this, which is why the Fluency Pathway covers responsible use rather than speed. Whether a training module can hold the line is a different question, and it is not one a bank can answer in a recruitment round.
What the 2027 intake actually tests The graduates who join UBS in 2027 will be among the first anywhere in European banking hired on these terms. They will arrive able to prove they use AI well. The job waiting for them is one where the tasks that used to teach the trade have already gone to software. That is the awkward shape of it.
A bank can screen for the skill that replaced the apprenticeship, and it still has to invent a replacement for the apprenticeship. The Fluency Pathway is UBS having a go at both problems with one programme. Banking has always claimed its apprenticeship was the point. Three years of grinding through models and decks, badly at first, until the judgement arrived.
UBS is now betting that judgement can be built a different way. Nobody finds out whether that worked until the class of 2027 is running deals, which is roughly a decade from now.
