aceGeneration AI: fears of ‘social divide’ unless all children learn computing skills
Abstract: In a Cambridge classroom, Joseph, 10, trained his AI model to discern between drawings of apples and drawings of smiles. “AI gets lots of things wrong,” he said, as it mistakenly identified a fruit as a face. He set about retraining it and, in a flash, he had it back on track – instinctively understanding the inner nature of artificial intelligence and machine learning in a way few adults do. His friends from the St Paul’s C of E primary school coding club tapped away to build their own AIs with similar dexterit…
By the publication date of 2026-01-05, The Guardian described a Cambridge classroom where Joseph, 10, from St Paul's C of E primary school coding club, trained an AI model to discern between drawings of apples and drawings of smiles. The model initially mistakenly identified a fruit as a face, and Joseph then retrained it to get it back on track while friends built their own AIs.
The episode matters because it shows young children encountering core machine learning concepts of training, error, and retraining through hands-on activity. What remains uncertain from the supplied excerpt is how widespread such opportunities are, whether they translate to lasting computing skills, and how the social divide referenced in the headline manifests beyond this single classroom.
- Joseph is 10 years old and attends St Paul's C of E primary school in Cambridge
- Activity took place in a school coding club
- Task was training an AI model to discern between drawings of apples and drawings of smiles
- Model initially made an error and was then retrained by the pupil
In a Cambridge primary school coding club, a 10-year-old was able to retrain his image classification model after an error, demonstrating practical learning of how machine learning models are corrected.
Sources
- JournalismThe Guardian2026-01-05
How should this claim be treated?
The debate