Pith. sign in

Reference change · event page

Reference changes · DOI

Generative AI Without Guardrails Can Harm Learning: Evidence from High School Mathemat- ics

Published notice on a work cited in the Pith corpus. Exact quotes below. No model judges whether any citation was load-bearing.

This page records that a citing paper's bibliography includes a work with a published notice. It is not a judgment on the citing paper.

Correction Crossref 5 open · 5 total · 0 disputed
DOI
10.1073/pnas.2422633122
Notice DOI
10.1073/pnas.2518204122
Event date
2025-08-20
Machine twin
JSON

01One-hop citing occurrences

Correction Open
Designing Human-GenAI Interaction for cMOOC Discussion Facilitation: Effects of a Collaborative AI-in-the-Loop Workflow on Social and Cognitive Presence

ref [11] · 2603.29285 · notice #527 · dispute

Raw extraction · bibliography line

Hamsa Bastani, Osbert Bastani, Alp Sungu, Haosen Ge, Özge Kabakcı, and Rei Mariman. 2025. Generative AI without guardrails can harm learning: Evidence from high school mathematics.Proceedings of the National Academy of Sciences 122, 26 (July 2025), e2422633122. doi:10.1073/pnas.2422633122
Correction Open
Agentivism: a learning theory for the age of artificial intelligence

ref [45] · 2604.07813 · notice #526 · dispute

Raw extraction · citation context

https://doi.org/10.1073/ pnas.1821936116. 20 [44] Bastani H, Bastani O, Sungu A, Ge H, Kabakci O, Mariman R. Generative AI Without Guardrails Can Harm Learning: Evidence from High School Mathemat- ics. Proceedings of the National Academy of Sciences of the United States of America. 2025;122(26):e2422633122. https://doi.org/10.1073/pnas.2422633122. [45] Liu M, Wu Z, Dai H, Su Y, Malik L, Liao J, et al. Enhancing self-directed learning and Python mastery through integration of a large language model and learning analytics dashboard. British Journal of Educational Technology. 2025;https:// doi.org/10.1111/bjet.70005. [46] Hu W, Gong R, Wu S, Li Y. A conversational agent based on contin- gent teaching model to support collaborative learning activities: impacts on
Correction Open
Faster Completion, Less Learning: Generative AI Reduced Study Time on Math Problems and the Knowledge They Build

ref [1] · 2605.21629 · notice #528 · dispute

Raw extraction · bibliography line

doi: 10.1073/pnas.2422633122. 12 Preprint. Under review. John Bound, Charles Brown, and Nancy Mathiowetz. Measurement error in survey data. InHandbook of econometrics, volume 5, pp. 3705–3843. Elsevier,

Parser render (TeX stripped for reading; raw above is the evidence)

doi: 10.1073/pnas.2422633122. 12 Preprint. Under review. John Bound, Charles Brown, and Nancy Mathiowetz. Measurement error in survey data. InHandbook of econometrics, volume 5, pp. 3705–3843. Elsevier

Correction Open
Two-player Alternate Uses Test: A Controlled Testbed for Interactive Human-AI and Human-Human Co-Creation

ref [3] · 2607.07522 · notice #530 · dispute

Raw extraction · bibliography line

Hamsa Bastani, Osbert Bastani, Alp Sungu, Haosen Ge, Özge Kabakçı, and Rei Mariman. 2025. Generative AI without guardrails can harm learning: Evidence from high school mathematics. Proceedings of the National Academy of Sciences 122, 26 (2025), e2422633122. https://doi.org/10.1073/pnas.2422633122

Status lifecycle on notices: open → disputed → (response attached on the notice page). Repaired counts matter as much as open counts. There is no “safe,” “invalid,” or “resolved” badge.