What AI fluency means for leaders making AI decisions

When a leadership team wants to become more advanced in AI, the conversation usually turns to tools within a few rounds of chat.

Which assistant, which licence, which training platform. Those are still fair questions, but none of them tells a leader whether the business is using AI well.

For a leader, AI fluency is the ability to ask the right questions and judge whether an AI investment is worth it. Asking them well requires a clear sense of what AI is good at and where it struggles.

In the EU, AI literacy is also a legal duty. Article 4 of the AI Act, as amended this year, asks organisations that use AI to take measures to support AI literacy among their staff. Literacy is the baseline for everyone who works with AI.

For a leadership team I'd call it AI fluency, the ability to govern AI, and the five questions below are how I'd test it.

1. What business problem does it solve?

I did some quick research on the executive AI programmes out there, and most of them start with value creation. Naming the business problem is the first step, before any technology decision. Then ask whether AI does it better than the way you work today, whether it fits how your teams already work, and whether you can test it small before you commit.

2. Which decisions does it affect?

This is where the collaboration between people and AI comes in. Which decisions does AI make on its own, where does it only suggest, and who signs off? The risk level of each decision tells you who should have the final say. A tool that drafts internal notes and a tool that decides which customers see an offer need very different conversations, even when they run on the same model.

A simple first step: list every place your organisation uses AI today, and next to each one write the decision it shapes, the people on the receiving end and the person who has the final say.

3. Can we trust it?

Trust is a big heading. It covers whether the information is accurate, whether the system is secure, whether the data is suitable for the purpose and you have permission to use it, whether you can explain how the system reaches a decision, and whether the results are consistent for different groups of customers.

4. Who is responsible when something goes wrong?

Decide at the start who answers for what the AI does and what it produces. Name who steps in, who is accountable and what the customer is told, and agree it before the first incident, while everyone still has time to think. In Grant Thornton's 2026 AI Impact Survey, leaders reported that 75% of their boards had approved major AI investments, and 48% had yet to set expectations for how that AI is governed.

5. How will we know it works?

Speed and cost are usually the first effects anyone reports, so they tend to become the only ones. Look further, at commercial growth, customer experience, brand trust and how your people's skills develop. If a tool makes a process twice as fast and customers complain twice as often, the speed figure on its own will wrongly tell you the project is a success.

Where to start this week

Take one AI project that is on your table now and put the five questions on the agenda of your next leadership meeting. Give each one ten minutes and one owner. Write the answers on a single page, including the ones that come back as "we don't know yet". Those gaps are your plan for the next quarter.

If you'd like to talk it through, bring the problem you're stuck on to a free 30-minute call.

The next two years

Tools change, models change and suppliers change. Over the next two years, the leadership teams that get lasting value from AI will be the ones that can answer these five questions for every AI project, whichever tools they use next.

← All insights