I wrote recently that no homework is safe now, and that the right response is to treat homework as practice, not assessment for grades.
Case in point: a 30-month study of 26,811 Chinese secondary students. When students adopted AI, homework scores rose 18% and homework time fell 30%—while exam scores fell 20% within six months.
But! Students who spent the same amount of time on homework as they did before AI didn’t see a drop in test scores. Learning takes work, and also, AI use doesn’t have to shortcut learning.
By contrast, AI users who spend as much time on homework as non-users achieve similar exam scores, even though their higher homework scores indicate that they use AI. These students aren ot differentially selected on prior achievement, suggesting that generative AI does not reduce their learning effciency.
The Generative AI Learning Penalty: Evidence from Chinese Secondary Education | SSRN