RCT: AI Tutoring Gains Persist a Week Later, AI Ghostwriting Gains Fade
emollick · x · 2026-10-10
A proctored randomized experiment (arXiv, Contractor & Reyes) had undergrads study an unfamiliar topic and write an essay with or without GPT-4o, then take unaided tests immediately and a week later. Key findings:
- AI access raised test scores, with a smaller gain persisting after one week
- Gains held for students using AI as a tutor ("augmentation") but faded for those who had AI write for them ("automation")
- Mechanisms: students shifted time from writing toward reading and searching, and reported greater learning enjoyment
- Effects were positive across all experiments
Ethan Mollick amplified the augmentation-vs-automation contrast.
Related event: RCT: AI Tutoring Boosts Learning, Copy-Paste Doesn't Stick(2 posts)→
More from Research
- NYU math prof: OpenAI's 722 AI-generated papers "destroyed" early-career mathematicians — tekbog · 2026-10-10
- Calico open-sources Cerberus DNA model in PyTorch: 786kb input, 8361 human tracks — anshulkundaje · 2026-10-10
- New preprint uses state space models for long-sequence genomic prediction, open-sources Cerberus PyTorch models — anshulkundaje · 2026-10-10
- MaCVi maritime computer vision workshop lands at WACV 2027 with 3D sonar and underwater benchmarks — HildeKuehne · 2026-10-10
- OpenAI pushes 722 math manuscripts to GitHub from a model nobody can use — thursdai_pod · 2026-10-10
- Neuroscience paper: AI model derives solution once deemed analytically intractable — scottleibrand · 2026-10-10