Breadcrumb
- Home
- AI Hub For Education
- Research Study Repository
- Evidence Explorer
- AI Tools
- ChatGPT / OpenAI GPT
The Repository
Search and Filter
ChatGPT / OpenAI GPT
Studies tagged with AI from their full text or abstract; see each study for details.
Showing 91 - 120 of 1083 studies
To Police Or To Guide: How Higher Education Computer Science Instructors Design And Implement Generative AI Policies
Xingjian (Lance) Gu, Wells Lucas Santo, James M. Zumel Dumlao, Barbara Jane Ericson. (07/2026). arXiv.
Instructors reported that generative AI misuse harms student learning through cognitive offloading, an illusion of competence, and social isolation, while also undermining instructors' ability to evaluate student work.
P-A.I.R.: A Structured AI-Replication Framework For Active Learning In Introductory Physics
Bilas Paul. (07/2026). arXiv.
Nearly all participants (97.4%) reported improved conceptual understanding, and 97.4% rated the AI-replicated practice problems as helpful.
Understanding How University Guidelines Address Privacy And Security Issues Of Generative AI In Academic Settings
Bei Yi Ng, Jiarui Li, Xinyuan Tong, Kevin Ye, Gauthami Yenne, Varun Chandrasekaran, Jingjie Li. (07/2026). arXiv.
Universities broadly recognize privacy risks involving sensitive student, research, and institutional data, but under-specify concrete adversarial behaviors and technical attack surfaces.
Student Evaluation Of Repeated AI Feedback Across A Semester Of Writing
Andres Karjus, Janika Leoste, Tiia Õun. (07/2026). arXiv.
Students found AI feedback helpful 57% of the time and actionable 53% of the time, though positive reactions declined over the semester.
Cost-Efficient Generative AI Summarization For Scalable Automated Essay Scoring In Educational Assessment
Haowei Hua. (07/2026). arXiv.
GPT-5 mini achieved the highest human-machine scoring agreement (QWK of 0.8435), slightly outperforming full GPT-5 (0.8350) and GPT-5 nano (0.8332).
Using AI-Based Learning Assistants In Higher Education: A Large-Scale Descriptive Analysis
Kristina Schaaff, Quintus Stierstorfer, Valerie Hekkel. (07/2026). arXiv.
During February 2025, 57.57% of distance learning students (44,035 out of 76,485) used the AI assistant Syntea.
How Well Does AI-Generated Feedback Work? Intrinsic And Extrinsic Evaluation Across More Than 20,000 Efl Essay Drafts
Steven Coyne, Diana Galvan-Sosa, Ryan Spring, Machi Shimmei, Michael Zock, Keisuke Sakaguchi, Kentaro Inui. (07/2026). arXiv.
Expert teacher ratings of feedback quality did not significantly align with students' thumbs-up or thumbs-down ratings (p=0.0625).
When Rubrics Change: Cross-Rubric Generalization For Critical Thinking Essay Scoring
Nischal Ashok Kumar, Payu Wittawatolarn, Sana Kang, Marisa C. Peczuh, Blair Lehman, Ryan Baker, Caitlin Mills, Sherry Lachman, Ruochen Sun, Andrew Lan. (07/2026). arXiv.
Rubric-agnostic traits improved scoring performance by 5.0% macro F1 when neither the target rubrics nor the target essays were seen during training.
Measuring How Students Rely On Generative AI In Academic Writing: Development And Multi-Source Validation Of The Generative AI Reliance Types Scale (Genai-Rts)
Shahin Hossain, Tukhbita Afroz Nawmi. (07/2026). arXiv.
A five-factor structure fit student writing reliance best, distinguishing deliberate use, critical evaluation, instrumental reliance, dependent reliance, and dialogic reliance.
Analyzing Curricular Pattern Complexity Using AI To Improve On-Time Graduation Rates
Lynn Vonderhaar, Juan Couder, Siri Siqveland, Omar Ochoa, James Pembridge. (07/2026). arXiv.
Claude Sonnet 4.6 was the only tested model that correctly calculated curricular complexity factors, while also providing interactive visualization tools and 83% reasonable revision suggestions.
A Comparative Analysis Of Institutional And Course Generative AI Policies Within Higher Education: Implications For Instruction In Computing Education
Amrita Ganguly, Aditya Johri, Nora McDonald, Areej Ali, Umama Dewan, Aayushi Hingle Collier. (07/2026). arXiv.
While 63% of universities encouraged generative AI use, half of computer science syllabi outright prohibited it, with only 41% allowing partial use.
Learning In Blocks: A Multi Agent Debate Assisted Personalized Adaptive Learning Framework For Language Learning
Nicy Scaria, Silvester John Joseph Kennedy, Deepak Subramani. (07/2026). arXiv.
Combining multi-agent debate scoring and recommendations with mastery-based progression and spaced review yielded significantly higher gains across grammar, vocabulary, and communication than feedback alone.
The Paternalistic Filter: Epistemic Injustice And Differential Refusal In LLM-Mediated History Education For Marginalized Romanian Students
Alexis Popovici, Andrei Ionascu, Adrian Marius Dumitran. (07/2026). arXiv.
Safety-aligned Kimi K2 refused 76.7% of requests from low-tier vocational students compared to 62.2% from ethnic Romanian and Roma profiles.
Faster Completion, Less Learning: Generative AI Reduced Study Time On Math Problems And The Knowledge They Build
Sina Rismanchian, Hasan Uzun, Jeffrey Matayoshi, Eric Cosyn, Eyad Kurd-Misto. (07/2026). arXiv.
Learning time on AI-susceptible math word problems dropped cumulatively by 31.3% for high schoolers and 26.9% for college students after ChatGPT's release, with no change in Grade 5.
LLM-Generated Design Problems For Assessing Higher-Order Thinking In Project-Based Learning
Ahmad D. Suleiman, Daqing Hou, Maliha Noushin Raida. (07/2026). arXiv.
Large language models generated high-quality design problems, with the reasoning model significantly outperforming the standard model on scenario realism and feasibility.
Generative AI In Higher Education Laboratory Learning: A Qualitative Case Study Of Epistemic Scaffolding And Assessment Boundaries
Matteo Tuveri, Alessandro Riggio. (07/2026). arXiv.
Students framed the AI tutor across five functions: interface interpreter, warrant organizer, report scaffold, unstable authority, and a resource leaving traceable marks in final reports.
How Annotation Trains Annotators: Competence Development In Social Influence Recognition
Maciej Markiewicz, Beata Bajcar, Wiktoria Mieleszczenko-Kowszewicz, Aleksander Szczęsny, Tomasz Adamczyk, Grzegorz Chodak, Karolina Ostrowska, Aleksandra Sawczuk, Jolanta Babiak, Jagoda Szklarczyk, Przemyslaw Kazienko. (07/2026). arXiv.
Annotators became significantly faster, reducing average annotation time from 6.56 to 5.86 minutes per text, while their self-perceived competence and confidence increased significantly.
Commenting With Copilot: A Taxonomy And Multi-Year Analysis Of Student Code-Generation Specifications
Nasser Giacaman, Valerio Terragni, Paul Denny, Viraj Kumar. (07/2026). arXiv.
Students mostly wrote natural-language comments describing intended behavior (77.0%), but shifted to procedural descriptions for multi-step blocks (94.3%) and loop iterations (57.3%).
Evidence-Decision-Feedback: Theory-Driven Adaptive Scaffolding For LLM Agents
Clayton Cohn, Siyuan Guo, Surya Rayala, Hanchen David Wang, Naveeduddin Mohammed, Umesh Timalsina, Shruti Jain, Angela Eeds, Menton Deweese, Pamela Joy Osborn Popp, Rebekah Stanton, Shakeera Walker, Meiyi Ma, Gautam Biswas. (07/2026). arXiv.
As student task mastery increased, the AI agent significantly reduced probing questions and increased concrete suggestions and limit-pushing prompts.
Learning Behavior Accounts For Background-Related Advantage In AI-Assisted Education
Jingwei Yi, Yueqi Xie, Jiyan He, Rui Ye, Junming Huang, Bin Zhu, Sean Rintel, Yu Xie, Xing Xie, Fangzhao Wu. (07/2026). arXiv.
Proactive and critical engagement with AI, such as attempting problems first or verifying answers, was associated with significantly higher exam scores than limited engagement.
When LLM Tutoring Responses Work: Evidence From Student Programming Conversations
Mohammad Fahim Abrar, Shayla Sharmin, Roghayeh Leila Barmaki. (07/2026). arXiv.
Across all interactions, response styles showed statistically significant but small differences, with verification feedback yielding the highest productive continuation rate (82.4%) and direct answers the lowest (62.7%).
Indirect And Direct AI Scaffolding For Computational Problem Posing: A Pilot Experience Report
Shayla Sharmin, Mohammad Fahim Abrar, Mohammad Al-Ratrout, Roghayeh Leila Barmaki. (07/2026). arXiv.
Both AI scaffolding approaches significantly improved problem quality, but direct scaffolding produced larger score gains than indirect scaffolding.
L2-Bench: An Evaluation Benchmark For Measuring LLM Capabilities In Second Language Education
James Edgell, Wm. Matthew Kennedy, Ben Knight, Danielle Carvalho, Martin Ku, Isaac Pattis. (07/2026). arXiv.
Across 1,000 tasks, Claude Opus 4.7 achieved the highest overall score at 85.5%, followed by GPT 5.4 at 84.1% and Gemini 3.1 Pro at 83.4%.
Flowcode: An AI-Powered Programming Environment For Scaffolding Iteration In Creative Computing Education
Tiffany Tseng, Liliana Hanem Seoror, Jeevika Adda, Meitalia Factor, Rona Darabi, Kiley R Matschke, Tiffany Fu, Annie Lin, Alekhya Maram, Arya Sinha. (07/2026). arXiv.
AI-generated flowcharts helped novice learners orient to multi-file codebases, navigate across languages, understand functional code relationships, and plan subsequent edits.
When AI Is Wrong On Purpose: How Students Respond To Buggy Genai Code
Victor-Alexandru Padurean, Kaitlin Riegel, Alkis Gotovos, Jyotika Mahapatra, Ahana Ghosh, Paul Denny, Juho Leinonen, James Prather, Adish Singla. (07/2026). arXiv.
Encountering deliberately injected bugs led students to edit code directly more often and resolve errors immediately (89.92% on first turns) compared to natural failures (51.16%).
AIED's Unfinished Mission: Centering Agency And Motivation In The Age Of Effortless Bypass
H. Chad Lane. (07/2026). arXiv.
Freely available generative AI creates an effortless bypass dilemma by allowing learners to outsource cognitive tasks, which risks undermining self-regulated learning and metacognitive habits.
Psychological Determinants Of Academic Integrity In The Use Of Generative AI In Higher Education
Ezgi Dagtekin, Ercan Erkalkan. (07/2026). arXiv.
Moral disengagement, peer influence, and policy ambiguity emerged across the reviewed literature as the most recurrent and behaviorally proximate determinants of generative AI misconduct.
Developing An LLM-Based Feedback System Grounded In Evidence-Centered Design To Support Physics Problem Solving
Holger Maus, Fabian Kieser, Stefan Petersen, Peter Wulff, Paul Tschisgale. (07/2026). arXiv.
Expert evaluation revealed that the AI-generated feedback contained errors in about 20% of cases, ranging from calculation errors to misclassifying correct alternative solutions.
Reflective Dialogue or Prompt Refinement? Effects of Tutor Scaffolding on Students' Independent LLM Use for Programming
Jérôme Brender, Laila El-Hamamsy, Kim Uittenhove, Aitor Perez, Patrick Jermann, Francesco Mondada, Engin Bumbacher. (07/2026). arXiv.
Students using the Socratic tutor achieved significantly greater learning gains by the final lab session than those using the prompt-refinement tutor, despite lower initial task performance.
Reflective Dialogue Or Prompt Refinement? Effects Of Tutor Scaffolding On Students' ndependent LLM Use For Programming
Jérôme Brender, Laila El-Hamamsy, Kim Uittenhove, Aitor Perez, Patrick Jermann, Francesco Mondada, Engin Bumbacher. (07/2026). arXiv.
Students using the Socratic tutor achieved significantly greater learning gains by the final lab session than those using the prompt-refinement tutor, despite having lower pre-test scores.
