#preprint
10 篇論文
Yeon Su Park, Chanyoung Park, Tae Soo Kim, Seungju Kim, Hyunwoo Kim, Daho Jung, Jinsung Yoon, Juho Kim
arXiv preprint
A benchmark builds simulated tutoring cases from qualitative education studies. Models were more successful at choosing a proposed strategy than at generating a response adapted to the particular situation.
Trystan Geoffre, Zheng Yuan, Thierry Geoffre
arXiv manuscript
A manuscript analyzes sequences of actions in a French grammar game to describe reasoning patterns and predict validation success. Its observational design leaves the learning effects of proposed scaffolding untested.
Zeyu Dong, Mikael Kwame Kuada, Asa Holley Buchakjian-Tweedy, Ibrahem Aljabea
arXiv preprint
A retrospective analysis compares tutoring-interaction features with constant score predictions. Small overall gains and uneven errors show why a support model needs more than a favorable average accuracy figure.
Rajeev Bhatt Ambati, Shashank Srivastava, Snigdha Chaturvedi
arXiv preprint
AdaTutor compares forms of assistance before selecting its next response. Tests with simulated learners suggest a benefit over fixed strategies, leaving real-student learning and classroom transfer to be established.
Daniel Domínguez Figaredo, Rafael Fernández De la Cruz
arXiv preprint
An optimization experiment produced higher automated scores but lower human ratings for instructional sequences. Its synthetic cases and constrained intervention make it a warning about evaluation design, not evidence of harm to actual students.
Sayantan Adak, Rima Hazra, Koyel Mukherjee, Digbalay Bose, Somnath Banerjee
arXiv preprint, version 2
A review examines how learner information, instructional choices and evaluation connect in AI tutoring. It highlights the need to inspect what students subsequently do, beyond the apparent quality of a conversation.
Xi Zhao, Xinyue Jiao, Zhen Xu
arXiv preprint
一份分析 3,041 份計算機科學答案的預印本發現，AI 的細分評分維度高度重疊，反饋中識別出的概念誤解也少於教師評語。它提出了一個實際問題：評估能否解釋學生錯在哪裏，而不僅是給出看似合理的分數？
Ming Li, Qin Xie, Ariunaa Enkhtur, Lilan Chen, Fei Cheng
arXiv preprint (work in progress)
Interviews with 42 undergraduates at universities in China, Japan, and Mongolia explored how students selected AI tools, chose interaction languages, and checked outputs. Participants described different routes from technical access to useful academic support, including switching languages when responses seemed inadequate. This work-in-progress preprint presents situated accounts rather than national comparisons or measured learning gains. It raises practical questions about evaluating an AI service in the languages students actually use.
Stylianos Psychias, Yingqiang Gao
arXiv preprint
AD4Edu proposes a benchmark and evaluation framework for describing visual information in university lectures. Four blind or low-vision raters assessed 400 description pairs, revealing a gap between compliance with the researchers’ rules and individual preferences, especially about length. This preprint offers tools for studying that gap, with small-sample and playback limitations. It does not demonstrate improved course learning or certify that a lecture meets accessibility requirements.
Po-Chin Chang, Nicholas Hogan, Aske Plaat, Michiel T. van der Meer
arXiv preprint, version 1
A high-school tutoring preprint examines adaptive prompt selection. Its interaction results invite closer study, but limited exercise data and nonsignificant outcome comparisons leave the learning advantage unresolved.