Applied foresight · childhood · education · governance

AI Tutors & Education 2035

A scenario exercise exploring how AI tutoring could shift educational inequality from simple access to quality of support, agency, representation and institutional capacity.

The question

How might widespread AI tutoring reshape children’s agency and educational inequality by 2035 — and what decisions remain robust across very different futures?

Decision lens: adoption alone is not the outcome. The important variables are meaningful access, quality of human support, cultural representation, accountability and the ability to contest automated decisions.

2035 scenario matrix

Weak governance
Strong governance
Limited school adoption
Educational Archipelago
Patchy experiments and invisible inequalities outside school.
Selective Adoption
AI is used only where evidence and oversight justify it.
Widespread school adoption
Subscription Ecosystem
Fast integration creates dependency before effective controls.
Shared Infrastructure
Broad use under transparent procurement, oversight and rights.

Robust actions

Evaluate alternatives

Compare supervised AI with human support and non-AI options, including retention without the tool and total cost.

Preserve options

Keep offline resources, portability, exit clauses and continuity paths so institutions are not locked into one provider.

Operationalise rights

Build child participation, accessibility testing and usable complaint mechanisms into procurement and pilots.

Limits

This is an exploratory foresight exercise, not a forecast or a systematic review. It was prepared independently and was not commissioned by UNICEF or any other organisation.