Why This Study Matters
While the debate around AI in education often focuses on whether AI helps or harms, this study reveals a more nuanced picture: how students use AI matters more than whether they use it. The distinction between substitution and complementary use has direct implications for how educators should structure AI-assisted learning.
Key Distinction: Substitution vs. Complementary Use
- Substitution: Students use LLMs to generate solutions to exercises. Result: more topics covered, but shallower understanding of each. Students may overestimate how much they have learned.
- Complementary: Students use LLMs to ask for explanations. Result: same topic coverage with deeper comprehension.
Equity Implications
The finding that LLMs widen the gap between low- and high-prior-knowledge students is especially concerning. Students who already know more are better positioned to use AI effectively, while those who need help most may be harmed by unsupervised AI access.
Bottom Line
AI in education is not inherently helpful or harmful — the outcome depends on usage patterns. Students who let AI do the thinking learn less; students who use AI to enhance their own thinking learn more. Educators need to actively shape how students interact with these tools.