Breadcrumb
- Home
- AI Hub For Education
- Research Study Repository
- Evidence Explorer
- AI Tools
- Custom-built Research Tool
The Repository
Search and Filter
Custom-built research tool
Studies tagged with AI from their full text or abstract; see each study for details.
Showing 451 - 480 of 531 studies
Metabook: A System to Automatically Generate Interactive AR Storybooks to Improve Children's Reading
Yibo Wang, Yuanyuan Mao, Shi-ting Ni, Wang Zeyu, Pan Hui. (11/2024). arXiv.
Metabook significantly increased children's interest in reading.
SBI-RAG: Enhancing Math Word Problem Solving for Students through Schema-Based Instruction and Retrieval-Augmented Generation
Prakhar Dixit, Tim Oates. (11/2024). arXiv.
The schema-based system achieved a higher reasoning score (0.588) on GSM8K problems compared to GPT-4 (0.491) and GPT-3.5 Turbo (0.290).
LLM-based Cognitive Models of Students with Misconceptions
Shashank Sonkar, Xinghe Chen, Naiming Liu, Richard G. Baraniuk, Mrinmaya Sachan. (10/2024). arXiv.
Models learned to replicate misconceptions with 400 to 800 examples, requiring fewer examples than the 2,000 needed to learn correct solutions.
FAIREDU: A Multiple Regression-Based Method for Enhancing Fairness in Machine Learning Models for Educational Applications
Nga Pham, Minh Kha Do, Tran Vu Dai, Pham Ngoc Hung, Anh Nguyen-Duc. (10/2024). arXiv.
FAIREDU achieved 35 wins out of 60 fairness comparisons against the prior benchmark LTDD, reducing disparate impact by up to 96.7%.
Emotionally Enriched Feedback Via Generative AI
Omar Alsaiari, Nilufar Baghaei, Hatim Lahza, Jason M. Lodge, Marie Boden, Hassan Khosravi. (10/2024). arXiv.
Students rated emotionally enriched AI feedback as significantly more useful than neutral feedback across both content authoring and peer review tasks.
How to Teach Programming in the AI Era? Using LLMs as a Teachable Agent for Debugging
Qianou Ma, Hua Shen, Kenneth R. Koedinger, Tongshuang Wu. (10/2024). arXiv.
Novice students using HypoCompass improved their pre-to-post test debugging scores by 11.7% and reduced completion time by 13.6%.
Designing Prompt Analytics Dashboards to Analyze Student-ChatGPT Interactions in EFL Writing
Minsun Kim, SeonGyeom Kim, Suyoun Lee, Yoosang Yoon, Junho Myung, Haneul Yoo, Hyunseung Lim, Jieun Han, Yoonsu Kim, So-Yeon Ahn, Juho Kim, Alice Oh, Hwajung Hong, Tak Yeon Lee. (10/2024). arXiv.
Teachers struggled with increased workloads when monitoring for subtle plagiarism and were concerned that ChatGPT paragraph generation diminished students' critical thinking.
Into the Unknown Unknowns: Engaged Human Learning through Participation in Language Model Agent Conversations
Yucheng Jiang, Yijia Shao, Dekun Ma, Sina J. Semnani, Monica S. Lam. (10/2024). arXiv.
In human evaluations, 70% of participants preferred Co-STORM over a search engine and 78% favored it over a retrieval-augmented generation chatbot.
Automated Feedback in Math Education: A Comparative Analysis of LLMs for Open-Ended Responses
Sami Baral, Eamon Worden, Wen-Chiang Lim, Zhuang Luo, Christopher Santorelli, Ashish Gurung, Neil Heffernan. (10/2024). arXiv.
The fine-tuned Mistral model (GOAT) outperformed SBERT-Canberra and zero-shot GPT-4 in predicting teacher scores, achieving an AUC of 0.697 and Kappa of 0.422.
Transforming Teachers' Roles and Agencies in the Era of Generative AI: Perceptions, Acceptance, Knowledge, and Practices
Xiaoming Zhai. (10/2024). arXiv.
Teachers progress through four distinct stages of generative AI engagement: Awareness (Observer), Exploration (Adopter), Integration (Collaborator), and Innovation (Creator/Innovator).
PromptHive: Bringing Subject Matter Experts Back to the Forefront with Collaborative Prompt Engineering for Educational Content Creation
Mohi Reza, Ioannis Anastasopoulos, Shreya Bhandari, Zachary A. Pardos. (10/2024). arXiv.
Hints generated using PromptHive produced student learning gains of 8.13%, which were not statistically significantly different from gains from human-authored hints (7.47%).
Hey AI Can You Grade My Essay?: Automatic Essay Grading
Maisha Maliha, Vishal Pramanik. (10/2024). arXiv.
The proposed Collaborative Deep Learning Network achieved the highest overall accuracy of 85.50 percent, outperforming BERT and other baseline deep learning models.
PAIGE: Examining Learning Outcomes and Experiences with Personalized AI-Generated Educational Podcasts
Tiffany D. Do, Usama Bin Shafqat, Elsie Ling, Nikhil Sarda. (10/2024). arXiv.
Students rated both generalized and personalized AI-generated podcasts as significantly more appealing and enjoyable than traditional textbook reading.
Mentigo: An Intelligent Agent for Mentoring Students in the Creative Problem Solving Process
Siyu Zha, Yujia Liu, Chengbo Zheng, Jiaqi XU, Fuze Yu, Jiangtao Gong, Yingqing XU. (09/2024). arXiv.
Students scored significantly higher on creative problem-solving learning reports when using Mentigo compared to a baseline system (13.43 vs. 10.74 out of total points).
BoilerTAI: A Platform for Enhancing Instruction Using Generative AI in Educational Forums
Anvit Sinha, Shruti Goyal, Zachary Sy, Rhianna Kuperus, Ethan Dickey, Andres Bejarano. (09/2024). arXiv.
Three out of four teaching assistants reported improved efficiency, and all four reported improved response quality using BoilerTAI.
Bio-Eng-LMM AI Assist chatbot: A Comprehensive Tool for Research and Education
Ali Forootani, Danial Esmaeili Aliabadi, Daniela Thraen. (09/2024). arXiv.
Bio-Eng-LMM successfully integrates conversational text, speech recognition, image understanding, and image generation within a single modular platform.
StuGPTViz: A Visual Analytics Approach to Understand Student-ChatGPT Interactions
Zixin Chen, Jiachen Wang, Meng Xia, Kento Shigyo, Dingdong Liu, Rong Zhang, Huamin Qu. (09/2024). arXiv.
Combining definition inquiries and follow-up questions with output restrictions formed an effective interaction pattern linked to high student scores.
"My Grade is Wrong!": A Contestable AI Framework for Interactive Feedback in Evaluating Student Essays
Shengxin Hong, Chang Cai, Sixuan Du, Haiyue Feng, Siyuan Liu, Xiuyi Fan. (09/2024). arXiv.
CAELF maintained high grading accuracy after student challenges, achieving 51.0% to 77.0% interaction accuracy across dimensions, while baseline models suffered roughly 30% drops.
LLaVA-docent: Instruction tuning with multimodal large language model to support art appreciation education‚ú©
Unggi Lee, Minji Jeon, Yunseo Lee, Gyuri Byun, Yoorim Son, Jaeyoon Shin, Hongkyu Ko, Hyeoncheol Kim. (09/2024). arXiv.
LLaVA-Docent progressed linearly through art appreciation stages by asking one question at a time, but heavily concentrated its questioning on personal interpretation.
Automating Human Tutor-Style Programming Feedback: Leveraging GPT-4 Tutor Model for Hint Generation and GPT-3.5 Student Model for Hint Validation
Tung Phung, Victor-Alexandru Padurean, Anjali Singh, Christopher Brooks, Jose Cambronero, Sumit Gulwani, Adish Singla, Gustavo Soares. (08/2024). arXiv.
The GPT4Hints-GPT3.5Val method achieved around 95% precision, matching human tutor quality while maintaining over 70% coverage across three real-world Python datasets.
Q-Module-Bot: A Generative AI-Based Question and Answer Bot for Module Teaching Support
Mia Allen, Usman Naeem, Sukhpal Singh Gill. (08/2024). IEEE.
Prototype and system evaluations confirm the tool is an effective, scalable cross-departmental system that provides real-time responses with source attribution.
New Curriculum, New Chance - Retrieval Augmented Generation for Lesson Planning in Ugandan Secondary Schools. Prototype Quality Evaluation.
Simon Kloker, Herbertson Bukoli, Twaha Kateete. (08/2024). arXiv.
Generated lesson plans achieved an average quality score between 75% and 80%, corresponding to a 'very good lesson plan' under the evaluation protocol.
Trust and ethical considerations in a multi-modal, explainable AI-driven chatbot tutoring system: The case of collaboratively solving Rubik's Cube
Kausik Lakkaraju, Vedant Khandelwal, Biplav Srivastava, Forest Agostinelli, Hengtao Tang, Prathamjeet Singh, Dezhi Wu, Matt Irvin, Ashish Kundu. (08/2024). arXiv.
Integrating intent classifiers and policy rules allows the chatbot to issue incremental warnings for abusive language and refuse queries that leak peer learning data.
RAMO: Retrieval-Augmented Generation for Enhancing MOOCs Recommendations
Jiarui Rao, Jionghao Lin. (07/2024). arXiv.
Unlike a traditional content-based and collaborative filtering system that failed on cold-start queries, the retrieval-augmented system provided relevant course suggestions for new users.
CourseAssist: Pedagogically Appropriate AI Tutor for Computer Science Education
Ty Feng, Sa Liu, Dipak Ghosal. (07/2024). arXiv.
CourseAssist achieved higher average instructor-evaluated scores than GPT-4 across usefulness (0.91 vs. 0.83), accuracy (0.92 vs. 0.71), and pedagogical appropriateness (1.00 vs. 0.82).
FairyLandAI: Personalized Fairy Tales utilizing ChatGPT and DALLE-3
Georgios Makridis, Athanasios Oikonomou, Vasileios Koukosa. (07/2024). arXiv.
User evaluations showed high engagement levels among young readers, reflected in increased reading time and positive feedback from children and parents.
Integrating AI in College Education: Positive yet Mixed Experiences with ChatGPT
Xinrui Song, Jiajin Zhang, Pingkun Yan, Juergen Hahn, Uwe Kruger, Hisham Mohamed, Ge Wang. (07/2024). arXiv.
About 65% of students used ChatGPT models, mainly for course content clarification, homework, and programming, but few felt highly confident in chatbot answers.
Generative AI Can Harm Learning
Hamsa Bastani, Osbert Bastani, Alp Sungu, Haosen Ge, Ozge Kabaci, Rei Mariman. (07/2024). SSRN.
While having access to GPT-4, student performance significantly improves by 48% with GPT Base and 127% with GPT Tutor.
COMET : "Cone of experience” enhanced large multimodal model for mathematical problem generation
Sannyuya Liu, Jintian Feng, Zongkai Yang, Yawei Luo, Qian Wan, Xiaoxuan Shen, Jianwen Sun. (07/2024). arXiv.
COMET achieved 33.84% accuracy across CMM12K math problems, outperforming open-source baselines including a model five times its size.
Building a Domain-specific Guardrail Model in Production
Mohammad Niknazar, Paul V Haley, Latha Ramanan, Sang Truong, Yedendra Shrinivasan, Ayan Kumar Bhowmick, Prasenjit Dey, Ashish Jagmohan, Hema Maheshwari, Shom Ponoth, Robert Smith, Aditya Vempaty, Nick Haber, Sanmi Koyejo, Sharad Sundararajan. (07/2024). arXiv.
A fine-tuned Mistral 7B model achieved an F1 score of 0.8056 on elementary classroom queries, outperforming fine-tuned Llama 2 13B, Gemma 2B, and Phi-2.
