# NZZ: AI Fluency Is Now Career-Relevant at UBS

> Quoted in NZZ am Sonntag on how companies test AI fluency, and the three questions I use to check it

- Date: 2026-09-06
- Source: https://romanoroth.com/en/publications/nzz-ki-fluency-ubs/
- Tags: AI, AI Fluency, Skills, AI Adoption, Leadership


## Summary

Companies have stopped asking whether employees can write a decent prompt. They want AI fluency, a confident and reflected way of working with AI. At UBS this has become career-relevant: according to research by NZZ am Sonntag, AI skills play a role in hiring, promotion and pay. Stefan Seiler, member of the group executive board and head of HR, puts it more carefully: "AI skills and practical experience are an important factor for future professional success." The bank opens an "AI power hour" in every employee's calendar each week and measures tool usage at an aggregated level.

Then comes the harder question: how do you measure a skill that has no curriculum yet? Marc Beierschoder, AI partner at Deloitte Switzerland, argues that neither login counts nor prompt volume say anything about real value, and that such metrics invite "AI theatre". One UBS employee told the paper he logs into Copilot in the evening so his score does not slip. Miriam Dachsel of Accenture Switzerland points at the gap between what leadership expects and what shows up in business results: licences make the office day easier, but the bigger efficiency gains only arrive once processes and training change with them.

The article also cites an Anthropic analysis of 9,830 Claude conversations (Anthropic Education Report: The AI Fluency Index). Only 30 percent of users set the terms of the collaboration up front, for example by asking the model to challenge their assumptions. Only 15.8 percent questioned the machine's reasoning. Checking facts was rarer still.

## My Quotes

For me, AI fluency does not begin with the software. It begins with the ability to reflect on your own work. I test that in conversations with three questions that look simple.

> "Which tasks do you handle differently today than six months ago, and how do you verify the result?"

The answer shows immediately whether someone reflects on their own way of working and develops it further.

> "Where do you deliberately not use AI, and why?"

Anyone who cannot name limits does not have any. Fluency also shows in deliberate restraint and in understanding where language models are weak and where they hallucinate.

With experienced professionals I shift the view from the individual to the system:

> "What have you built that other people are still using today?"

That last question is the one that separates personal productivity from something the organisation actually keeps. At Zühlke we look at whether people manage to carry their AI workflows over to their team.

---

**[Read the full article on NZZ](https://www.nzz.ch/wirtschaft/ki-kompetenz-wird-immer-wichtiger-bei-der-ubs-haengt-sogar-die-karriere-davon-ab-ld.10018898)**