GenAICanHarmLearning anonymized dataset and code repository¶
Product or Programme · All products and programmes
Evidence · no claims cited
Description¶
A GitHub release by Bastani Hamsa et al. containing anonymized practice- and exam-performance data, student–GPT-tutor conversations, covariates, and R, Python, and Stata analysis code from a high-school mathematics generative-AI learning RCT.
Components¶
- Anonymized dataset and code repository for the generative AI learning RCT: The authors released anonymized data and analysis code at https://github.com/obastani/GenAICanHarmLearning. Data include "student performance on practice and exam questions for all sessions, as well as student-, class-, and grader-level covariates", plus anonymized, time-stamped student conversations with the GPT tutors. Code was written in R, Python, and Stata. (Bastani Hamsa et al. (2025))
Claims¶
Related Products and Programmes¶
- GenAI-Learning-Outcomes GitHub repository
- GenAI Education Reddit Corpus and 17-Theme Taxonomy
- LLM-queries
- Open online dataset and statistical analyses of the mutual-modelling emotions study
Key Sources¶
- Bastani Hamsa, Bastani Osbert, Sungu Alp, Ge Haosen, Kabakcı Özge, Mariman Rei. (2025). Generative AI without guardrails can harm learning: Evidence from high school mathematics. Proceedings of the National Academy of Sciences of the United States of America. https://doi.org/10.1073/pnas.2422633122
Page metadata
| Field | Value |
|---|---|
type |
product |
id |
genaicanharmlearning-anonymized-dataset-and-code-repository |
title |
GenAICanHarmLearning anonymized dataset and code repository |
description |
A GitHub release by Bastani Hamsa et al. |
status |
review |
product_kind |
dataset |
generated |
by process:settle-candidates on 2026-10-12 |
trust |
unverified — no one has confirmed this page's content |
Frontmatter is the page's machine-readable record — see the Schema & Guide.