My feed is full of AI papers, workshops and conferences. I welcome the interest. I want more of it to translate into practical competence and better professional work.
For me, the most valuable learning comes from pushing AI to the limits of what I can get it to do. I deliberately try demanding tasks that I am not yet sure it can handle.
Sometimes I return to the same problem for days, exploring the tools and trying different approaches. When I finally get it working, I understand much more about what is possible, where the limits lie and what I need to check.
AI use is already widespread: Gallup found that 60% of US public-school teachers used it for work in 2024/25. In Thomson Reuters’ 2026 report across legal, tax and other professional services, 40% of respondents reported organisational use of generative AI; 15% reported agentic AI.
“Using AI” can mean writing a report, preparing a presentation or creating an image. It can also involve research, data analysis and work across several documents.
A chat can draw on uploaded files and connected tools. Tools such as Codex or Claude Cowork can carry out several steps across files and applications, including browser or computer operation where enabled. Projects can bring related tasks, subtasks and findings together.
Consider curriculum mapping. AI can help break a curriculum into subject areas, learning outcomes and competencies, and map them against labour-market evidence. Bring in evidence of teachers’ competencies and actual classroom practice as well. The mapping can reveal gaps and contradictions between what is intended, what teachers are equipped to deliver and what is actually taught and assessed.
The curriculum on paper and the curriculum in the classroom may differ. Making that difference visible is useful; deciding what to change requires professional judgement and dialogue with educators.
Model choice and access matter too. Free and paid plans can differ in available models, tools and usage limits. Compare a fast model with one designed for deeper reasoning on the same difficult task. Assess the quality of the result.
I judge AI’s value by whether it helps me produce work that has been examined more thoroughly and rests on stronger evidence.
In my view, books, conferences and workshops do little to build that judgement unless people also work seriously with the tools.
For education and consulting, I would put sustained practice at the centre of AI learning: suitable tools, demanding tasks and informed feedback.
After the next AI workshop, ask: what can you now do better with AI, and how do you know?
Sources & further reading
Adapted for the website from the original LinkedIn post. Views and professional reflections are the author’s.
Read the original post and discussion on LinkedIn
