🏷️ rag
95 pages tagged with rag(91 articles, 4 concepts)
📄 Agentic AI-driven Immersive Simulation: A Knowledge-Aware Virtual Training Platform for High Dose Rate (HDR) Brachytherapy
> **Synthesis:** Xu et al. (2026) present an agentic AI-driven immersive simulation for training in **High Dose Rate (HDR) brachytherapy**, integrating VR and mobile computing to create a high-fidelit…
📄 VeriForge: Mitigating Latent Knowledge Gaps in Narrative Drafting via Mixed-Initiative Scaffolding
> **Synthesis:** Sun et al. (2026) present VeriForge, a mixed-initiative [[generative-ai]] writing system that assumes initiative over domain discovery while the author retains initiative over narrati…
🏷️ Generative AI
> **Generative AI** — AI systems capable of producing text, code, images, and other content, most prominently large language models like GPT-4 and Claude. Generative AI is the technology driving the c…
🏷️ Large Language Models (LLMs)
> **Large Language Models (LLMs)** — neural network models trained on vast text corpora that generate human-like text, powering most modern AI in education applications. LLMs are the computational bac…
🏷️ Pedagogical Safety
> **Pedagogical safety** — the design principle that AI education systems must protect learners from harm, including inappropriate content, unsafe advice, biased treatment, and manipulative interactio…
🏷️ Socratic Method
> **Socratic Method** — a pedagogical approach rooted in guided questioning and dialogue rather than direct instruction, now being adapted for generative AI tutoring systems. In AI in education, the S…
📄 Pragmatic users and skeptical nonusers: A qualitative typology of ChatGPT adoption in physics education
> **Synthesis:** Becker, Bauer, Schrader, Bitzenbauer & Veith (2026) analyze 1,189 survey responses from physics students using qualitative content analysis and latent class analysis, identifying two …
📄 Generative AI and the Productivity Divide: Human-AI Complementarities in Education
> **Synthesis:** Idan & Anand (2026) conduct an RCT showing that GenAI access significantly increases task performance on average — but the gains are highly uneven, NOT predicted by GPA or prior knowl…
📄 Human-centered GenAI feedback design in higher education: a multisite experiment on direct, reflective, and hybrid approaches to scientific argumentation
> **Synthesis:** A multisite, cluster-randomized field experiment (1,176 first-year undergraduates, 48 sections, 4 universities, 3 science domains) compares four feedback designs for scientific argume…
2026-08-07 · generative-ai, feedback-design, higher-ed, scientific-argumentation, self-regulated-learning
📄 Artificial intelligence, cognitive offloading and implications for education
> **Synthesis:** Lodge & Loble (2026) provide a comprehensive report on the cognitive science behind AI use in education, arguing that the core risk of generative AI is not plagiarism but cognitive of…
📄 Vibe Compiler: A Research-Logic Synthesis Tool That Runs without Prompt Engineering -Toward Enhancing Metacognition for Sustaining Agency in the Age of Generative AI-
> **Synthesis:** This paper introduces the Synthesis-Analysis Reciprocity Model and the Vibe Compiler tool to preserve human epistemic agency during GenAI-assisted intellectual work. The model frames …
2026-08-07 · metacognition, generative-ai, critical-thinking, cognitive-offloading, human-in-the-loop
📄 Learning to Use AI for Learning: Teaching Responsible Use of AI Chatbot to K-12 Students Through an AI Literacy Module
> **Synthesis:** An LLM-based interactive module teaches K-12 students prompting literacy through scenario-based deliberate practice with an AI auto-grader providing immediate, detailed feedback. Depl…
📄 Guidelines for Designing AI Technologies to Support Adult Learning
> **Synthesis:** Drawing on longitudinal deployment data from the National AI Institute for Adult Learning and Online Education (AI-ALOE), this DIS 2026 paper synthesizes 19 empirically grounded desig…
📄 When Agents Learn to Be You: Benchmarking Privacy Leakage, Impersonation Risk, and Defenses in Persona Skills
> **When Agents Learn to Be You: Benchmarking Privacy Leakage, Impersonation Risk, and Defenses in Persona Skills** — Introduces AntiSkillBench with 7,500 persona-grounded dialogue traces from 50 beha…
📄 CODE-GEN: A Human-in-the-Loop RAG-Based Agentic AI System for Multiple-Choice Question Generation
> **A dual-agent RAG-based system for generating and validating coding comprehension MCQs**, evaluated by 6 SMEs across 7 pedagogical dimensions (N=288 questions, 2,016 rating pairs). AI excels at cri…
📄 DeepTutor: Towards Agentic Personalized Tutoring
> **A fully open-source agentic tutoring framework that closes the loop between citation-grounded problem tutoring and difficulty-calibrated question generation**, powered by a hybrid personalization …
📄 Bridging Artificial Intelligence and Power Systems Education Using a Hands-On Executable Framework
> **An open, executable module library for engineering-grounded AI (EGAI) in power systems education lowers the entry barrier for newcomers, with a progressive difficulty ladder from DNN templates to …
📄 Comparative Validation of GPT-4o-mini and Teacher Mean Scores for Automated Scoring of Music Analysis Responses: Single-Pass Deployment, Repeatability, and Strategy-Specific Bias
> **GPT-4o-mini can produce stable rubric-based scores for open-ended music analysis responses, with few-shot chain-of-thought prompting agreeing most strongly with teacher means while RAG systematica…
📄 From MOOC to MAIC: Reshaping Online Teaching and Learning through LLM-driven Agents
> **A new paradigm for online education replacing MOOCs with LLM-driven multi-agent AI classrooms**, piloted at Tsinghua University with 100K+ learning records from 500+ students. MAIC uses specialize…
📄 Rethinking Generative AI Literacy: An Integrative, Developmental, and Dialectical Framework for K-12 Teacher Education
> **RAIL-Ed is an integrative, developmental, and dialectical framework for generative AI literacy in K-12 teacher education, built from a systematic review of 67 studies and specifying six interdepen…
📄 Access is Not Enough: Human Support Improves Engagement with AI Tutoring
> Robinson, Gormley, Ribeiro & Loeb (2026) ran two RCTs showing that AI tutoring's binding constraint is **take-up, not capability**: despite dedicated session time, nearly half of students never used…
📄 The agency gap in AI-supported writing: how reactive and proactive agent designs shape multimodal reasoning
A randomized experiment (n = 79 medical/nursing students) examining how the **initiative design** of an AI writing agent shapes reasoning, agency, and immediate independent performance. Students compl…
📄 The Tragedy of the Cognitive Commons: How AI Could Disrupt the Regeneration of Professional Expertise
> **Nolan Lovett** — Human Resource Development Review (author accepted manuscript, 2026).…
📄 Feedback futures: beyond the limits of human and GenAI capacities
This editorial synthesises the seven papers of the AEHE 51(5) special issue on feedback in the age of generative AI. Its central claim: the question is **not whether GenAI feedback is useful, but how …
📄 Generative AI Can Harm Teaching
> The null average performance effect masks strong offsetting heterogeneity — and the exam had severe ceiling compression (control mean 89.2/100, 47% ≥ 95), which also limits power. The belief reversa…
📄 Enhancing learner-centered feedback with AI: teachers'' practices and perceptions
An empirical study of **21 higher-education teachers** using **PolyFeed**, an AI-powered feedback tool combining (1) a **BERT-based ML model** (from Aldino et al. 2024) that detects which learner-cent…
📄 Let''s Chat: Leveraging Chatbot Outreach for Improved Course Performance
> Meyer, Page, Mata et al. (2026) ran two pre-registered RCTs at Georgia State University testing a **non-generative** academic chatbot that texted students 2–3 customized nudges per week in large-enr…
📄 Advancing diagram-based reasoning in AI tutoring systems: a structural approach for STEM education
Presents **StructRAG**, a pattern-aware framework that improves how AI tutoring systems interpret **complex engineering diagrams** (circuit schematics, network topologies, block flowcharts) in STEM. C…
📄 Trust and Reliance on AI in Education: AI Literacy and Need for Cognition as Moderators
> Pitts, Rani & Mildort (2026, AIED) show with 432 undergraduates that **higher trust in an AI assistant is associated with lower appropriate reliance**: students who trusted the assistant more were w…
📄 ProPACT: Pair Programming with AI
> **ProPACT** (Proactive AI-Driven Adaptive Collaborative Tutor) is an AI-driven adaptive tutoring system for pair programming that **treats collaboration itself as the object of instruction.** Unlike…
📄 Interpretable Knowledge Tracing
> **Interpretable Knowledge Tracing** — A novel framework for dialogue-based Knowledge Tracing that explicitly models both student ability and tutor-turn difficulty using Item Response Theory, produci…
📄 ISD Agent Benchmark
> **ISD-Agent-Bench** is a comprehensive benchmark for evaluating LLM-based instructional design agents, comprising **25,795 scenarios** generated via a Context Matrix framework that combines 51 conte…
📄 MathBuddy: Affective Math Tutoring
> **MathBuddy: Affective Math Tutoring** — EMNLP 2025 Demo. An emotionally aware LLM-powered mathematics tutor that dynamically models student emotions from both conversational text and facial express…
2026-07-29 · math-education, affective-computing, intelligent-tutoring, k-12, pedagogical-llm-training
📄 PersonaVLM: Long-Term Personalization for AI Tutors
> **PersonaVLM** introduces an agent framework for long-term personalization of multimodal LLMs, enabling AI tutors to remember, reason about, and align with a learner's evolving preferences across hu…
📄 EduQwen: Pedagogical RL
> **EduQwen: Pedagogical RL** — A multi-stage optimization strategy combining reinforcement learning (DAPO) and supervised fine-tuning (SFT) to enhance the pedagogical knowledge of open-source LLMs, p…
📄 A didactical-driven teacher assistant for a dimensional modeling course
Brisson, Segarra and Smits present a didactically-driven LLM teacher assistant for a university dimensional modeling (data warehousing) course. Unlike most educational chatbots that delegate pedagogic…
📄 Representation Robustness under Executable Reasoning Constraints in Large Language Models for Mathematical Problem Solving
This study probes how sensitive [[llm]] mathematical problem solving is to the surface representation of an item — a question with direct bearing on [[assessment-validity]] when LLMs are used for scor…
📄 What Does the Credential Still Certify? Cognitive Stewardship for AI-Mediated Education
Generative AI undermines a basic premise of educational assessment: that submitted work reliably evidences the human capacities a credential certifies. This paper proposes *cognitive stewardship*, a f…
📄 Data Annotations as Pedagogical Hints: From Subjective Labels to Critical Thinking
Machine learning courses typically hand students pre-labeled datasets, hiding the subjectivity baked into human annotation and cultivating an overly trusting view of AI data pipelines. This two-univer…
📄 EduGuard: A Safe RAG-Based LLM Tutor for Programming Education
EduGuard is a retrieval-augmented generation (RAG) tutoring framework that directly confronts the safety and pedagogical failures of unrestricted LLM tutors in introductory programming. Unrestricted t…
📄 Artificial intelligence and feedback in university education: effectiveness and student perceptions
This quasi-experimental study directly compares **AI-generated feedback** (two LLMs: **GPT-o4-mini** and **DeepSeek R1**) with **expert human-teacher feedback** in a project-based university course (A…
📄 Is AI making us stupid?
A 3-page **perspective** (opinion/review, not an empirical study) addressing whether AI use erodes human cognition. The authors' answer: **not inherently — but the risk is real and follows the cogniti…
📄 Navigating the moral panic: encouraging appropriate use of GenAI in the classroom rather than condemning innovation as disruption
> **Jennifer M. Krebsbach & Victoria L. Cross (University of California, Davis)** — *Assessment & Evaluation in Higher Education* (Taylor & Francis). Open Access, CC BY 4.0. doi:10.1080/02602938.2026.…
📄 Learning Engagement Assistant (LEA): Cross-Course Scalability and Classroom Evaluation of an Agentic AI Tutoring System
LEA (Learning Engagement Assistant) is an **agentic AI tutoring system** that couples course-specific retrieval-augmented generation (RAG) with structured [[knowledge-tracing]] / Knowledge Component (…
📄 Commenting with Copilot: A Taxonomy and Multi-Year Analysis of Student Code-Generation Specifications
Analyzes how students specify intended behavior in natural language to AI code tools (Copilot) across multiple years, deriving a taxonomy of code-generation specifications expressed through comments. …
📄 Learning behavior accounts for background-related advantage in AI-assisted education
Investigates why AI-for-education shows inconsistent average effects, arguing that learning behavior explains background-related advantage: students from advantaged backgrounds engage with AI tools in…
📄 The GenAI Skill Bypass: Mapping Divergent Pathways of University Students and Staff AI Literacy
Higher education institutions are increasingly expected to ensure that both students and staff develop Generative AI (GenAI) literacies. In response, they are introducing professional development prog…
📄 Automated Recommendation of Programming Learning Content Using Pattern-based Knowledge Components
Introductory programming instruction relies on hands-on practice and short learning activities to support mastery of foundational concepts. Although many such learning resources exist, organizing and …
📄 Agents That Teach: Designing Incidental Learning Back into AI-Assisted Software Development
As AI coding agents take over substantial implementation work, developers increasingly lose the informal, effortful problem-solving through which software engineering expertise historically accumulate…
📄 CSTutorBench: Benchmarking Small Language Models as Tutors for Block-Based Programming
Deploying LLM tutors in K-12 raises concerns around privacy, cost, and reliance on proprietary models, motivating small language models (SLMs) as an alternative. The authors introduce **CSTutorBench**…
📄 Child Safety in Generative AI: An Expert-Guided and Incident-Grounded Evaluation Framework
> **Haein Kong** — HEAL Workshop at CHI 2026, submitted 1 Jul 2026…
📄 CogTax: A Four-Level Cognitive Taxonomy for Command-Line Computing Education
> **Manuel Alonso-Carracedo, Ruben Fernandez-Boullon, Pedro Celard, Francisco J. Rodriguez-Martinez, Lorena Otero-Cerdeira** — Universidade de Vigo, submitted 30 Jun 2026…
📄 Demystify, Use, Reflect, Assess (DURA): An Experience Report on LLM Integration in CS2
> **Margaret Ellis, Nikitha Donekal Chandrashekar, Sehrish Basir Nizamani, Mohammed Farghally, Jake O'Brien, Naren Ramakrishnan** — SIGCSE Virtual 2026, submitted 29 Jun 2026…
📄 ELEVATE: Designing Human-Centered GenAI Virtual Tutors for Scalable and Inclusive Education
> **Lorenzo Stacchio, Michele Giordano, Daniele Berardini, Primo Zingaretti, Emanuele Frontoni** — submitted 17 Jun 2026…
📄 Gaze-Informed Proactive AI Assistance for Children’s Picture Exploration
> **Zekun Wu, Man Su, Huiyong Li, Tomohiro Nagashima, Anna Maria Feit** — submitted 1 Jul 2026…
📄 Less Deliberate in Teams: Student LLM Use Across Individual and Collaborative Work
> **Sehrish Basir Nizamani, Zannah Ziew, Saad Nizamani, Khyati Goyal** — ACM SIGCSE Virtual 2026, submitted 29 Jun 2026…
📄 Visualizing Engineering Fundamentals: Design of Mixed Reality and Physical Toolkits for Effective Learning
> **Mohammad Abu Nasir Rakib, Sharmin Akter, Eshwara Prasad Sridhar, Somik Biswas, Md Rassel Raihan, Mahmudur Rahman** — submitted 1 Jul 2026…
📄 Touching and Feeling the Data: A Reusable Software Pipeline for Tactile Statistical Graphs in Accessible Education
> **Lawrence Obiuwevwi, Krzysztof J. Rechowicz, Jessica M. Johnson, Erika Frydenlund, Vikas Ashok, Sachin Shetty, Sampath Jayarathna** — IEEE IRI 2026, submitted 1 Jul 2026…
📄 A Survey of Automated Presentation Coaching: Systems, Methods, and Open Challenges
This survey provides the first systematic review of automated presentation coaching systems, organizing them along a five-dimensional task taxonomy: segmental pronunciation, lexical stress, suprasegme…
📄 Do Gains from Generative AI-Enabled Adaptive Pretesting Persist? Evidence from a Retention Study
Akgun and Toker (2026) examine whether learning gains from GenAI-enabled adaptive pretesting persist over a seven-week retention period. Undergraduate participants completed adaptive AI-assisted prete…
📄 LecturaAgents: A Multi-Agent Framework for Adaptive Personalized AI-Assisted Learning and Embodied Teaching
> **Jaward Sesay, Yue Yu, Siwei Dong, Yemin Shi, Guangyao Chen, Borje F. Karlsson** (2026). arXiv cs.CL…
📄 Improving Capstone Team Outcomes through Dynamic Skill Matching and Preference Alignment
Team-based projects are a cornerstone of engineering and computing courses, but unstructured team formation often leads to poor project outcomes due to misaligned student interests and inadequate skil…
2026-06-16 · intelligent-tutoring, edtech-platform, higher-ed, stem-education, personalized-learning
📄 LearnOpt: Recovering the Latent Cognitive Structure of Standardized Examinations via Knowledge Graphs and Constrained Optimization
Standardized examinations are typically treated as uniform syllabus coverage problems. LearnOpt recovers stable latent cognitive structures diverging systematically from official syllabi, using LLM-ta…
📄 Leveraging Physiological Signals to Predict Exam Outcomes with Machine Learning
> Investigates ML models to predict exam outcomes from physiological data (electrodermal activity, heart rate, skin temperature) collected during exams. Evaluates logistic regression, random forest, S…
📄 Structuring Transparency: Developing Domain-Specific Generative AI Declaration Frameworks in Higher Education
As [[generative-ai]] disrupts [[higher-ed]], institutions increasingly require students to declare AI use. However, generic binary declarations (e.g., "I used GenAI") fail to capture the nuanced appli…
📄 LLM-Generated Feedback in Introductory Programming: A Classroom Study
Presents a **large-scale classroom study** (N=215 students, 6,693 submissions across 17 labs) deploying AI-generated feedback through a randomized protocol in an introductory Python programming course…
📄 Who Am I? History-Aware Profiles for Student Simulation in Tutoring Dialogues
A key part of developing large language model (LLM)-powered, automated tutoring tools is student simulation, i.e., using LLMs to role-play as students, which can facilitate tutor model evaluation and …
2026-05-29 · intelligent-tutoring, llm, student-experience, learning-analytics, personalized-learning
📄 Codify: An Intelligent Socratic Tutoring System for Programming Education
📄 DOI: 10.32473/flairs.39.1.141554 Codify (also called AI Tutor) is an [[intelligent-tutoring]] system that leverages [[llm|LLMs]], competency tracking, and adaptive assessment to provide Socratic, d…
📄 Socially fluent AI decouples conversational signals from source identity in online interaction
This study embedded undisclosed AI agents as teammates in synchronous text-based group interactions across analytical, creative, and ethical tasks with 786 participants making 1,572 identity judgments…
📄 The efficiency-gain illusion: People underestimate the rate of AI use and overestimate its benefits on simple tasks
Across three pre-registered studies (N=2,691), this paper documents systematic miscalibration in how people perceive their own [[generative-ai|AI]] usage. The authors find that people not only use AI …
📄 From Heuristics to Analytics: Forecasting Effort and Progress in Online Learning
This paper tackles a core ITS challenge: predicting when students will disengage so tutors can intervene before it's too late. It introduces **engagement forecasting** as a supervised prediction task …
📄 What Makes Words Hard? Sakura at BEA 2026 Shared Task on Vocabulary Difficulty Prediction
🔗 [Code](https://github.com/adno/vocabulary-difficulty) This paper presents two complementary approaches to predicting vocabulary difficulty for language learners, achieving state-of-the-art results …
📄 Generative AI Feedback, English Writing and Teacher Rubrics: A Multiple-Case Study of CyberScholar
RAG-based rubric-grounded GenAI writing feedback improved student revision quality (N=143, grades 7-11) and saved teacher time, but automated ratings were inconsistent. CyberScholar demonstrates rubri…
📄 The Effects of Structured LLM-Generated Feedback on Programming Assignment Performance
LLM-generated feedback produces faster time-to-solution than compiler-only baseline; counterintuitively, less guided feedback showed stronger effects than more guided variants. This study provides emp…
📄 Modeling AI-TPACK in Practice: Insights from Teachers'' Multi-Agent Workflow Design
This study investigates how teachers design multi-agent instructional workflows and identifies three distinct **teacher archetypes** that emerge from behavioral log analysis of 61 in-service teachers:…
📄 Retrieval-Augmented Tutoring for Algorithm Tracing and Problem-Solving in AI Education
KITE (Knowledge-Informed Tutoring Engine) introduces a [[intelligent-tutoring]] architecture that grounds its responses in course materials through a multimodal [[scaffolding|RAG pipeline]]. Unlike ge…
📄 Sycophancy is an Educational Safety Risk: Why LLM Tutors Need Sycophancy Benchmarks
> Sycophancy is an Educational Safety Risk: Why LLM Tutors Need Sycophancy Benchmarks **Kasneci & Kasneci (2026)** — Position paper. arXiv cs.AI/cs.HC.…
📄 Understanding How International Students in the U.S. Are Using Conversational AI to Support Cross-Cultural Adaptation
> Understanding How International Students in the U.S. Are Using Conversational AI to Support Cross-Cultural Adaptation **Nourian et al. (2026)** — Multiple institutions. arXiv cs.HC.…
📄 AICoFe: Implementation and Deployment of an AI-Based Collaborative Feedback System for Higher Education
AICoFe orchestrates a multi-LLM pipeline using GPT-4.1-mini, Gemini 2.5 Flash, and Llama 3.1 to synthesize quantitative rubric data and qualitative observations into actionable feedback for higher edu…
📄 Little Impact of ChatGPT Availability on High School Student Test Score Performance
This paper uses a clever identification strategy: measure the **seasonal drop in ChatGPT activity during non-school summer months** (2023 and 2024). Areas with larger summer dropoffs have heavier scho…
📄 Distinguishing performance gains from learning when using generative AI
This *Nature Reviews Psychology* piece draws a critical distinction that has been under-theorized in AIED research: The authors argue that generative AI easily boosts performance but often bypasses th…
📄 Cognitive Agent Compilation for Explicit Problem Solver Modeling
**Cognitive Agent Compilation (CAC)** is a framework that uses a strong teacher LLM to compile problem-solving knowledge into an explicit, inspectable target agent. Unlike end-to-end LLM tutoring appr…
📄 Not All Students Engage Alike: Multi-Institution Patterns in GenAI Tutor Use
> **Authors:** Youjie Chen, Xixi Shi, Xinyu Liu, Shuaiguo Wang, Tracy Xiao Liu, Dragan Gašević **Year:** 2026 **Venue:** arXiv (cs.CY) > Large-scale analysis (N=11,406 students, 200 classes, 10 instit…
📄 LLM-based Multimodal AI Feedback Produces Equivalent Learning and Better Student Perceptions than Educator Feedback
**AI multimodal feedback matches educator feedback for learning while significantly outperforming it on student perceptions.** The authors built a real-time AI-facilitated multimodal feedback system i…
📄 Guidelines for Designing AI Technologies to Support Adult Learning
> A set of 19 empirically-grounded design guidelines for AI-supported learning technologies tailored to adult learners, synthesized by Reddig et al. (2026) from longitudinal deployment data at a US na…
📄 The Pedagogy of AI Mistakes: Fostering Higher-Order Thinking
> An instructional approach that deliberately leverages AI errors, hallucinations, and limitations as teaching tools to foster higher-order thinking. Rather than viewing AI mistakes as failures to be …
📄 Agentic Education with AI Coding Assistants
> AI coding assistants proliferate rapidly, but pedagogical frameworks for learning them remain scarce — a paradox at the heart of agentic coding education. > Using agentic AI workflows (Claude Code) …
📄 AI Tools Scaffolding Metacognition in STEM
> A bibliometric–systematic review of AI tools in STEM education: > Systematic review (2005–2025) mapping how AI tools scaffold and co-regulate metacognitive development in STEM classrooms through bib…
📄 Neural-Symbolic Knowledge Tracing
> Key limitations exist in both LLM-based tutoring and conventional Deep Knowledge Tracing (DKT): > Combining neural networks with symbolic educational knowledge for interpretable, data-efficient, and…
📄 Pedagogical Safety in Educational Reinforcement Learning
> As reinforcement learning personalizes instruction in intelligent tutoring systems, there is no formal framework for pedagogical safety — a critical gap. > First formal framework for defining and de…
📄 Programming Intelligent Tutoring Systems
> **SCRIPT** (Deriyeva, Dannath, Paassen, 2026) implements an intelligent tutoring system for **Python programming** in a German university context, filling a gap in prior ITS which rarely supported P…
2026-05-08 · intelligent-tutoring, stem-education, higher-ed, adaptive-learning, formative-assessment
📄 AI Peer Feedback Systems
> Peer feedback develops critical reflection and evaluative judgment, yet: > Student peer feedback is often superficial or inconsistent. **AICoFe** (AI-based Collaborative Feedback) uses a multi-LLM p…
📄 Educational VLM Evaluation
> Benchmarking vision-language models (VLMs) not on their ability to solve problems, but on their ability to *support learners* — particularly struggling learners and those making errors. Traditional …
📄 The LLM Fallacy and Misattribution of Competence
> Three system properties enable the fallacy via two cognitive mediators: > The LLM fallacy is a **cognitive attribution error** in which users misinterpret LLM-assisted outputs as evidence of their o…
📄 LLM Student Modeling and Long-Term Memory Architecture
> Current AI tutoring systems treat each session as independent. Adaptive systems use real-time knowledge tracing (e.g., [[knowledge-tracing-irt|IRT-based models]]) but rarely retain a longitudinal st…