I am not buying this study. They should have had a control group of students who did the work from home without the AI tutor. The underlying causal mechanism could be as simple and banal as “fewer distractions lead to better results”. This needs a lot of replication and a much better design.
I am not buying this study. They should have had a control group of students who did the work from home without the AI tutor. The underlying causal mechanism could be as simple and banal as “fewer distractions lead to better results”. This needs a lot of replication and a much better design.
I also wrote a piece about this paper! https://garyliang.substack.com/p/ai-tutoring-outperforms-active-learning