Study Links Heavy AI Use to 20% Drop in Exam Scores for Chinese Students

2 min read
Source: Futurism
Study Links Heavy AI Use to 20% Drop in Exam Scores for Chinese Students
Photo: Futurism
TL;DR

A large-scale study in China reveals that while AI tools boost homework scores and speed, they significantly undermine long-term learning and exam performance.

Key points

  • Researchers from Stockholm University and the University of Hong Kong tracked nearly 27,000 Chinese students aged 12 to 18.
  • Approximately 80% of the participants used AI models such as DeepSeek and ByteDance’s Doubao.
  • AI users saw an 18% increase in homework scores and a 30% reduction in time spent on assignments.
  • Despite higher homework grades, AI users scored 20% lower on monthly exams compared to non-users.
  • The study found the most significant learning losses occurred in social sciences, followed by STEM and language courses.

Background

This development follows earlier concerns about AI’s impact on critical thinking and memory, including an MIT study showing reduced brain activity among students using AI for writing. It also aligns with broader trends of grade inflation in AI-vulnerable college courses since the release of ChatGPT.

How outlets are covering it

Futurism emphasizes the 'ominous' nature of these findings, highlighting the disconnect between high homework scores and low exam results as a sign of AI being used as a cratch. The secondary source, standard-journal.com, provides no substantive content on the topic, consisting only of website interface code, and thus offers no perspective on the study.

Why it matters

The findings challenge the conventional assumption that high homework grades predict exam success, suggesting that AI may be fostering superficial learning rather than deep understanding. This has significant implications for how schools assess student knowledge and the potential long-term cognitive effects of relying on AI for academic tasks.

What to watch

Educators and policymakers may need to reconsider assessment methods, potentially shifting toward in-person exams or other measures that are harder to game with AI. Further research may be needed to understand the long-term cognitive effects of AI use in education.

Share this article

Want the full story? Read the original reporting

Read on Futurism