Forum: How will we know if AI safeguards in schools work?

The Ministry of Education’s Forum reply on AI use in schools is reassuring in terms of design (Schools guide students to use AI purposefully for learning, Sept 18).

The curriculum for younger children remains print-first, AI is introduced progressively as they get older, and teachers may deliberately and appropriately withhold its use. The Learning Assistant in the Student Learning Space requires teacher activation, allowing a cap on interactions, and prompts students to think instead of simply giving answers.

These are sensible safeguards. However, they describe intentions rather than outcomes.

In a written parliamentary reply in February, Education Minister Desmond Lee said MOE had no Singapore-centric data on whether AI use erodes cognitive abilities. Also noting international research that poorly structured use can hinder basic recall and understanding of simple concepts, the minister said MOE was studying the impact of AI use on students’ learning (MOE studying the impact of AI on learning; holds off on such tools for younger pupils: Desmond Lee, Feb 25).

As guard rails against over-reliance are only as good as our ability to test whether they work, I would ask MOE to share the scope and timeline of its study and when findings will be published, what findings would lead it to slow or pause the roll-out, and how the guard rails, and the readiness for AI use at each age, have been or will be independently evaluated.

This question serves an aim MOE and I share – that AI should support our children’s learning without replacing the thinking and effort required for learning.

Daphne Anne Rodrigues