During a recent Excel and AI workshop, I ran a group project and let each team choose whatever digital tools they wanted to produce their final report. Partway through, one participant said something that has stayed with me since: the AI did the report better than I do.

It was said almost in passing. But it is exactly the kind of comment that should make anyone who teaches — or works with — data pay attention.

That comment connects directly to a concern I raised in one of my own assignments for my Master of Education in Learning Design and Technology. Over-reliance on AI is already a widely recognised worry. What gets far less attention is its effect on the human and social side of adult learning. Adult learners and educators tend to be confident AI users, which is precisely what makes the slow erosion of critical thinking, independent reasoning, and meaningful human interaction so easy to miss. Recent research backs this up: over-dependence on AI in higher education is increasingly linked to reduced cognitive engagement and weakened analytical skills, with ethical awareness and institutional safeguards acting as important moderating factors (Abubakar et al., 2025).

This connects directly to Vygotsky's sociocultural theory, which places human interaction at the centre of deep learning. When AI starts mediating too much of that interaction, learning risks becoming transactional rather than transformative — you get an output, but not necessarily an understanding. Regulatory frameworks such as the EU AI Act are a signal of where this is heading: institutions will increasingly be held accountable for how AI is deployed in learning environments. For educators, staying informed on this is not optional. It is part of the job now.

AI helps us design better. But how we use the data, who we include, and what actually reaches our learners — those decisions still sit with us, not the tool.

The same logic applies in finance

The same logic applies directly to data analysts, and to anyone in finance producing a report that someone else will act on. Professionals are responsible for what they report — not the AI. When a number is wrong, it is the analyst who faces the consequences: the internal policies, the regulations, the decisions the business makes on the back of that report. AI does not sit in that meeting. You do.

That is why "verify, don't trust" is not a slogan I use lightly in my workshops — it is the whole point. AI is a genuinely useful partner for first drafts, formula suggestions, and spotting patterns you might otherwise miss. But the judgement, the accountability, and the final check before anything reaches a board pack or a client report has to stay human.

So, learning is still important. Not despite AI being capable — because of it.

Reference: Abubakar, A. et al. (2025). Research on AI over-dependence and cognitive engagement in higher education.

If your team is using AI tools day-to-day but isn't sure how to build a verification habit around them, get in touch about the Excel + AI workshop.