- Who it helps
- Educators evaluating tutoring support and unaided assessment
- Availability
- October 22 public preprint describing 153 students; this is evidence from one course setting.
Early access helped homework, without a clear exam gain
Randomized staggered access produced a reported 2.56-point homework increase, with no statistically significant exam difference. Homework performance therefore cannot establish an improvement in later unaided problem solving.
Usage associations need a different interpretation
After everyone received access, greater chat use was associated with lower exam scores. That association does not prove chat caused the decline. Feedback on proof attempts did not show the same negative association.
What you can do
- Read the primary description and its stated limits before planning an evaluation. Open reference ↗
Where the evidence stops
Abstract checked; data and analyses not reproduced. Related earlier papers from the same study are not independent replications.
Sources read. Public source text checked November 14, 2026 UTC. No third-party project, model inference or full workflow executed.
Sources & further reading
Paper · paper
AI tutoring in proof-based mathematics ↗
October 22 abstract distinguishes randomized early access from later observational usage associations.
Paper authors · page checked 2026-11-14 · Published 2026-10-22
Web · analysis
Proof homework and cognitive offloading ↗
November 5 research summary read. Independent explanation of the study, without replication or measured current attention.
Ask Poppy · page checked 2026-11-14
Coverage note: Abstract checked; data and analyses not reproduced. Related earlier papers from the same study are not independent replications.
The record over time (2)
Dated source checks and published explanations. An observation date can differ from the original release date.
- · Source recheck
A mathematics-course study examines AI chat and feedback on students’ proof attempts, including later exam performance.
- · Added to Current
A mathematics-course study examines AI chat and feedback on students’ proof attempts, including later exam performance.