Can AI harm learning?

This season Libby Hills from the Jacobs Foundation and AI researcher Owen Henkel continue to speak with leading researchers, practitioners and educators on the Ed-Technical podcast series about the cutting edge of AI in education. They will break down complex AI concepts into non-technical insights to better understand what the research says and help educators sift the useful insights from the AI hype. 

In this short, Libby and Owen discuss a recent paper that has generated interest and discussion called ‘Generative AI Can Harm Learning’. The paper presents the findings from a thought-provoking study of nearly 1,000 students in Turkey. The study tested the effects of giving students access to two different versions of GPT-4 while studying math: one was essentially ChatGPT and the other was a version of GPT-4 that had been tailored for tutoring with a thin prompt wrapper – so it didn’t just give students the answer.

The main finding (that the title is based on) is that access to generic ChatGPT had a negative effect on students’ math test results, versus the control group who studied with no access to a chatbot. Not everyone agrees that the results justify the somewhat dramatic title, or that the title reflects the most interesting findings from the study. Listen in to see what Libby and Owen think.

The ‘Generative AI Can Harm Learning’ paper can be found here.