When people ask which human skills will matter as AI spreads, I often hear lists of things AI supposedly cannot do. That seems a poor way to plan a career. AI will keep improving. I would rather ask: what skills help us work with it to get a better result?
Think about the whole piece of work
Imagine a service analyst trying to understand a rise in customer complaints. Using an approved AI tool and anonymised messages, she groups the complaints and gets a draft summary. It suggests that late deliveries are the main problem.
She checks a sample of the original messages and speaks to the service team. The deliveries are often on time; customers are frustrated because updates arrive too late. She proposes a clearer update, then watches whether repeat contacts fall over the next month.
AI helped her get through the messages. Asking the right question, checking the evidence and listening to colleagues helped her solve the right problem. The value came from how she put those skills and the tool together.
The skills I would develop
I would learn enough about AI to know what to ask of it and where it tends to go wrong. Alongside that, I would practise framing problems, testing evidence, understanding the people affected and following through on a decision. These skills become more useful together.
That fits the direction of current research. The World Economic Forum’s 2025 employer survey expects technology skills to grow fastest while also naming analytical thinking and leadership as important. A June 2026 OECD brief points to growing importance for using and interpreting data. Neither report gives us a permanent list of safe skills.
AI will improve at some of the checking and conversation too. I expect the work to change again. That is another reason to learn how to direct, assess and apply the tools, instead of building an identity around tasks they cannot yet do.
Try this on your own work
Choose a recurring piece of work. Where could an approved AI tool help? What would you need to check or learn from other people before acting on its output? Name one result you could measure—fewer corrections, a faster decision or better customer feedback. Try it, and see whether the whole piece of work improves.
If you need a starting point, use the 20-minute task audit. The free ADAPTOR framework can then help you turn what you learn into a personal SWOT and next steps.
Sources and further reading
- World Economic Forum, Future of Jobs Report 2025: Skills outlook (7 January 2025)
- OECD, AI and skills: What we know so far (5 June 2026)