A new study tracking 27,000 students in China found that pupils who used artificial intelligence tools saw higher homework scores over time, but performed worse than peers on exams taken without AI assistance, according to research covered by The Economist on August 18.

The study, conducted by David Stromberg of Stockholm University with Victor Lei and Wu Yanhui of the University of Hong Kong, followed pupils aged 12 to 18. Around 80% reported using AI models such as Doubao and DeepSeek, while the remaining 20% formed a control group. AI users saw homework scores rise by 18% across all subjects over six months, but scored 20% below non-users on exams without AI access.

The findings echo a 2024 University of Pennsylvania study where students using AI performed better in short-term practice but saw no advantage on a subsequent closed-book test.

Broader data shows widespread adoption: a Chegg survey found 80% of undergraduates in wealthy countries use AI, with figures at 94% in Britain and 93% in Germany. Commenters attributed the exam gap to students copying AI answers without deep engagement, while Brookings cautioned overreliance could hinder cognitive and social skill development. The full study had not been independently verified at publication.