Beyond the "Average" Learner: How AI Could Help Personalise Education, and Where It Can't

The problem with “average”

In the 1950s, the US Air Force measured more than 4,000 pilots on ten body dimensions to design a cockpit for the “average” pilot. Not one pilot was average on all ten. As Todd Rose describes in The End of Average (2016), designing for the average can mean designing for no one.

Learning works in a similar way. Any class contains learners who are ahead in some areas and still developing in others. Teachers know this well, and good teaching already adapts to it. The challenge is doing so for 20 or 30 learners at once.

Why this matters now

The skills the world needs are changing. The World Economic Forum’s Future of Jobs Report 2025 lists big data specialists, fintech engineers and AI and machine learning specialists among the fastest-growing roles to 2030. Green jobs such as renewable energy engineers also feature strongly.

That does not make broad knowledge less important. Specialists still need strong foundations in reading, maths and reasoning. It does suggest learners benefit from time to go deep in areas where they show strength and interest.

What AI can and cannot do

UNESCO’s Guidance for Generative AI in Education and Research (2023) recommends a human-centred approach: AI should support teachers and learners, not replace them. Used well, AI tools can help to:

• Adjust pace and practice, offering extra examples or stretch questions as learners need them.
• Reduce barriers, for example through text-to-speech, speech-to-text and reading support for learners with dyslexia.
• Surface patterns, helping teachers notice where a learner is engaged or stuck.

AI is less suited to building motivation, judging what a learner is ready for, or giving the encouragement and relationships that keep teenagers engaged.

It also matters how much guidance learners get. Cognitive load theory (Sweller, 1988) shows that working memory is limited. Learners new to a topic do best with clear, structured guidance, not unguided discovery (Kirschner, Sweller & Clark, 2006). Personalisation works best when it adds structure, not when it removes it.

How WeLearn uses technology

At WeLearn Academy, technology supports, but does not replace, our coaches. Learners aged 11–18 follow personalised pathways with weekly mentorship. We use AI tools to help with practice, research and accessibility, and we teach learners to use AI thoughtfully and honestly in their own work.