A two-year study conducted by a law professor compared three approaches to AI use in education: a total ban, unrestricted use without instructions, and structured training. According to The Decoder, the group without access to AI performed worse in both years.
The researcher writes that he was mistaken, as he previously considered unregulated AI application to be more harmful than beneficial. The source provides no details on the number of students, specific assignments, or how performance was measured.
The practical implication of the report is a shift from simple bans to evaluating specific rules and learning formats. However, this is an inference drawn from a brief summary, not a confirmed conclusion from a full scientific report.
editorial commentary
Why it matters
A likely consequence is increased interest in rules and training for working with AI instead of a universal ban. The next observable signal will be the publication of the full study with methodology and data. Significant uncertainty remains because currently only a brief summary from one source is available.