AI tools accelerate learning or stunt cognition

Proponents argue that specific AI design features drive measurable learning gains. In a study of 770 Taiwanese high schoolers, an LLM-powered tutor that dynamically adjusted problem difficulty raised scores by 0.15 standard deviations on proctored exams (Edtech Insiders). Middlebury College students who used AI as a tutor to explain concepts saw test scores rise by 0.27 standard deviations, with gains persisting a week later (Edtech Insiders). These tools succeed when they force students to engage with material rather than providing immediate answers.

Critics maintain that accessible general-purpose chatbots damage learning by encouraging students to bypass effort. At a public charter school in Austin, students abandoned the tool Class Companion after realizing they could copy-paste AI-generated evidence into their responses to regain points (Jenny Anderson from How to Be Brave). An analysis of 3.2 million ALEKS interactions found that students spent less time on problems AI could easily solve after ChatGPT’s release and subsequently performed worse on proctored assessments.

The evidence indicates that the utility of AI depends on whether the tool imposes mastery requirements or allows for passive output. While adaptive systems can improve performance, general-purpose chatbots often lead students to prioritize efficiency over the struggle required for retention.

• Google’s Gemini Guided Learning raised math scores by 0.26 standard deviations for 1,400 students in Sierra Leone over eight weeks (Edtech Insiders).
• A randomized trial of 90 Taiwanese tenth graders found that using ChatGPT’s Study Mode for Socratic dialogue improved critical thinking and scientific argumentation.