# Students Using AI Are Scoring Worse in School

> Global OECD data reveals that AI shortcuts hurt test scores, but trained, intentional use gives learners an edge.
- Title: Students Using AI Are Scoring Worse in School · The Verge
- Summary: Global OECD data reveals that AI shortcuts hurt test scores, but trained, intentional use gives learners an edge. AI only improves learning when used…
- Keywords: education, artificial intelligence, pisa, learning, edtech, technology, culture, Students, Using, Scoring, Worse, School
- Source: The Verge — https://theverge.com/ai-artificial-intelligence/991956/student-ai-use-scores-oecd-pisa
- Author: Dominic Preston
- Published: 2026-09-09T12:05:01+00:00
- Read time: 1 min
- Topics: education, artificial intelligence, pisa, learning, edtech, technology, culture
## AI study tools backfire
A global OECD study of over 760,000 students found that those using AI for schoolwork generally perform worse on standardized tests than those who abstain.
## Shortcuts cause the biggest drop
Using AI to draft essays or summarize required reading leads to steep performance drops. Both daily power users and rare annual users scored worst compared to moderate weekly users.
## Critical evaluation flips the script
When students receive training to critically assess AI accuracy and use it for general guidance, their test scores actually surpass those of non-users.
## Protect the cognitive struggle
True learning requires active mental friction. Technology aids growth when it deepens thought, but undermines development when it short-circuits the struggle of understanding.

> Where technology enables or enhances that cognitive struggle, students will advance.
## Curiosity drives adoption worldwide
Daily AI users exhibit the highest curiosity levels, but usage varies wildly across borders, spanning over 95 percent of students in Vietnam to just 60 percent in Japan.
## Key takeaway

AI only improves learning when used intentionally as a critical study aid rather than a shortcut that replaces mental effort.