AIEd Research Weekly Update #3November 7, 2026

A weekly brief from the AI in Education Knowledge Base, on what the research published in the past week means for instructors, developers, researchers, and administrators.

Since the last issue, 159 new studies, eight new concept pages, two new FAQs, and three new resources have joined the knowledge base, which now holds 2,085 article summaries, 230 concept pages, 44 FAQs, and 40 resources. Search or use AI to chat with the knowledge base.

Contents

• Main highlights and questions
• New FAQs
• New concepts
• New resources
• For instructors
• For developers
• For researchers
• For administrators
• Also added this week
• Site notes
• Follow the update

Main highlights and questions

• Can AI feedback replace a teacher's feedback? AI-generated feedback and teacher feedback improved essay revisions equally in a randomized trial with 312 first-year writers. Teacher feedback raised motivation more because students felt connected to the person giving it. AI can carry a revision pass; the teacher still carries the relationship.
• Can a supportive chatbot substitute for a peer? A supportive chatbot performed the same as journaling in reducing loneliness in a randomized trial with 296 first-year students; a randomly assigned peer reduced it. The bot expressed more empathy while students gave less back, so measure loneliness itself.
• Does a short ethics module change AI judgment? A four-week information ethics module raised students' scores on AI-related ethical judgment scenarios in a trial that randomized 16 classes. About 60% of the gain held four weeks later, and the heaviest AI users gained most. Case-based teaching shows promise.
• Does more AI assistance improve creative work? Four randomized experiments with 728 adults found that when AI did more of the work, judges rated the resulting innovation proposals lower, mostly because participants did less of the cognitive work. More assistance came with weaker work.
• Can an AI-authored game teach better than a lecture? A mystery role-playing game beat the same instructor's lecture on legal knowledge and reasoning in a randomized trial with 84 law students. The game was built with generative AI; the comparison was game versus lecture, so the result supports game-based learning and leaves chatbot use untested.
• Do student-performance reports change teaching? Teachers new to an online math tutoring system were randomly prompted to open a learning analytics dashboard report of their students' performance on the first assignment. Opening it changed neither the share starting the next assignment nor students' test-score gains, though next-assignment completion rose, mostly in smaller classes.
• Does matching human graders make an AI grader safe? An audit of six AI models grading 300 short answers found all marked identical work higher when told the student was high-achieving, by 0.28 to 0.75 points on a ten-point scale. Agreement can hide a bias tied to the student's prior-attainment record; withhold the record.
• Why is AI-generated language so hard to understand? AI text overuses rare word combinations, denominal verbs, inanimate subjects, reduced object relatives, stacked nouns, and long dependencies because post-training for math, code, and reasoning rewards compressed answers over clear explanations. The implication for education is that AI-written explanations, feedback, and reading material can be harder for students to process even when readability scores look fine, a problem for reading education that formulas miss. [Editorial note: As a reminder, this newsletter and the AIEd Knowledge Base are AI-generated. The site developer gave extensive editorial guidance to improve readability and usefulness; after that guidance, the AI agent applied the Humanizer skill and followed an explicit instruction to write affirmatively and avoid the "not X, but Y" construction.]

New FAQs

Two new answers joined the FAQ this week.

How can AI help students with reading and reading comprehension?

The reading answer treats reading-level labels as rough guesses: a few real readers' ratings beat a formula, and an AI told to pretend to be a student cannot tell you how hard a text will feel. Use AI to find, simplify, and gloss readings, then try the result with students and keep the original beside it.

How does AI bias and unfairness show up in education, and what can we do about it?

The bias answer asks for error rates for the groups you teach, since an average can hide a gap and the failure may sit in scoring, tutoring, dashboards, or detection. Put group reporting and an appeal route into procurement, and fix the pipeline itself.

New concepts

Eight concept pages joined the knowledge base this week.

Reading education

Reading education now has a page that separates four things which often share one label: helping a reader, assessing reading, adapting a text, and measuring difficulty. It cites a study where 193 students rated 3,954 sentences; 72.8% felt easy or very easy even when Flesch scores said otherwise, so a grade-level label is a weak guide to how a particular reader will experience a text.

Reflection

Reflection has a new page that separates it from the in-the-moment monitoring called metacognition. The first design question is whether the student wrote the reflection. In a 302-adult experiment, reflection prompts doubled time with feedback and cut practice by 40%, leaving lower mastery, 65% against 79%, so timing matters as much as the prompt itself.

Automation bias

Automation bias names the human habit that can turn a correct AI recommendation into a wrong decision: accepting an answer you would have questioned if you had reached it yourself. It collects the places where that happens in education, from essay scores and early alerts to proctoring, and shows how a review step that exists only on paper fails to catch it.

Civic and political education

Civic and political education maps a gap in the evidence and synthesizes the field. An analysis of 310 AI-generated lesson plans found 90% of activities asked only for recall, comprehension, or basic application; 94% had almost no multicultural content; and no study in the corpus measured civic learning at all.

Libraries

Libraries treats the library as a place, a service, and a group of professionals. A review of 29 studies finds adoption concentrated in reference and cataloguing, while the Public AI Libraries pilot makes shared, staffed access the equity argument.

AI avatars and synthetic instructors

AI avatars map the roles a synthetic human can occupy, from recorded lecturer and tutor to simulated patient and examiner. The evidence is uneven, and a realistic face tells you little about learning; disclosure and trust matter as much as the model.

Sensor and gaze analytics

Sensor and gaze analytics covers eye tracking, brain and heart signals, and camera posture as ways to infer attention and engagement, and cognitive load. The field is better at watching learners than at changing instruction, and the page treats privacy, bias, and the gap between a body signal and a mental state as central.

Information and media literacy

Information and media literacy treats the two as one instructional problem: information literacy asks whether a source deserves belief, while media literacy asks who made a message and why. Generative AI collapses both questions because fluent text hides its origin; the page finds little evidence on a combined curriculum.

New resources

Three resources joined the resources library this week.

LearnHarness

LearnHarness is a free, open-source desktop app in which AI agents work out what a learner already knows, map the ideas needed to reach a goal, teach one reasoning step at a time, and adapt practice.

Learning Commons

Learning Commons publishes open building blocks for curriculum products: a map of standards and curricula, tools for scoring generated text, and ready-made AI workflows for teachers.

Public AI Libraries

Public AI Libraries is a pilot that places a shared AI workstation in US public libraries, with chat, image and video generation, a coding workspace, an educational guide, and a private account for each patron.

For instructors

AI feedback matches a teacher on revisions while motivation lags

AI and teacher feedback improved the same essays by about the same amount in a 312-writer randomized trial. Teacher feedback did more for motivation because the student felt connected to the person giving it.

Keep a human voice around the feedback. Pair the generated comment with a short instructor response, or use AI for the first pass and the teacher for the conversation about the work.

See related: AI feedback at scale

One in five essays was written entirely by the chatbot

A study of 77 college writers logged every prompt, response, and paste. Fifteen, or one in five, submitted chatbot-written essays, and most still felt the essay was theirs.

Ask for the intermediate work: the prompts, the revisions, and a short note on what changed and why. Make the handover visible while there is still time to teach.

See related: Redesigning assessment for the AI era

Secondary students use AI almost universally, and lower achievers use it more

A year-long study of 4,062 German secondary students found nearly universal AI use: 95% used it, and lower achievers used it more to finish tasks. That pattern predicted less effort later.

That makes framing AI use and self-regulated learning part of the curriculum, revisited across the year. Ask students what the tool is for on this task, and give lower-achieving students a supported route that begins with understanding the work.

See related: Reducing over-reliance on AI

A supportive chatbot matched journaling while a peer reduced loneliness

In a two-week randomized trial with 296 first-year students, a supportive chatbot performed the same as journaling in reducing loneliness, while a randomly assigned peer reduced it. The bot showed more empathy while students gave less back.

A bot can serve as a useful first contact while a peer or counselor remains available. Build that human route into any student support service you recommend, and measure loneliness itself.

See related: AI, anxiety and wellbeing

Treat an AI simulation as supervised rehearsal

A review of 14 studies found nursing students valued repeatable, low-pressure AI simulation and reported more confidence, especially in communication and clinical reasoning. Realism, body language, and response lag remained weak.

Brief students on the limits, require them to check generated content, and debrief afterward. Keep human interaction where nonverbal communication or sustained empathy is the point of the exercise.

See related: Nursing education

Capstone code tripled while comments held steady

Ten semesters of 260 capstone projects in computer science education show median code rising from about 3,400 lines to 10,000, while comments stayed near 8%. Commit counts rose; individual commits barely changed.

Assess the design decisions, the explanation of the code, and the student's ability to change it. The artifact can now grow faster than the learning it is meant to show.

See related: Redesigning assessment for the AI era

For developers

Without the figure, an AI tutor should refuse

Six image-capable AI models tackled 376 middle-school math items whose figures carried key information. With the figure, the best solved just over half; without it, they refused almost every time.

Treat a refusal as a missing-context signal. A score or profile update produced without the figure measures a different task.

See related: Math education

A writing tutor that asks questions and leaves the writing to students

MAIKE asks questions and diagnoses arguments while leaving essay text to the student. Among 65 secondary students, improvement matched a control chatbot, 40% to 41%, and MAIKE users spent more turns on ideas; none declined.

Measure whether an idea discussed in the chat becomes a revision. A writing education tool that only improves the final text can bypass the reasoning the assignment exists to build.

See related: Writing instruction with AI

Zoom and highlight the slide while the narration runs

GuideSlide narrates a slide deck and uses visualization to zoom to the part being discussed. In a study of 24 people watching on phones, workload fell, comprehension rose from 5.5 to 6.5 out of 9, and 21 of 22 preferred it.

The advantage a day later was not statistically significant, so the study supports attention while watching; durable learning remains untested. Build the cueing into the narrative itself.

See related: Designing AI into learning

A fine-tuned detector can de-identify student text without the name bias

Five tools were tested on 13,613 student-text files. A fine-tuned GPT-4o-mini found about 96% of identifying details at roughly one sixth of Azure's per-token price; the Microsoft tools missed African and Asian names more often.

Training on ten transcripts worked on a new set, so the approach travels. Budget for the second check when precision matters, and report the groups your tool misses.

See related: Institutional AI policy

Video analysis reached human-level agreement, with a human still in the loop

GAIVA transcribes, labels, and interprets classroom video for learning analytics. On five recordings, its labels agreed with human reviewers about as closely as two reviewers did, and it surfaced a kind of participation the original theory did not name.

Treat the model as a coding partner with a traceable prompt and a human decision at each stage. The model also surfaces patterns a human reviewer might not have looked for.

See related: AI-assisted educational research

For researchers

A large average effect can still mean no reliable benefit

A review of four controlled studies with 420 computing and engineering students found a large average improvement, while the studies disagreed sharply and the likely range included no benefit. The evidence was rated very uncertain.

Treat a large average from four dissimilar studies as a hypothesis about your course. Pilot the personalized learning where it will run, and say plainly how uncertain the evidence is.

See related: What personalized learning means

An AI grader that knows the learner's record marks the same work differently

Six AI models graded the same 300 short answers with a true and a swapped note about prior attainment. Every model marked identical work higher under the high label, by 0.28 to 0.75 points on a ten-point scale.

Student context is acting like a bias here. Report the context effect beside the agreement figure, and test grading without the student's record before calling a model reliable.

See related: AI grading fairness and reliability

A fair-looking tutor can hide a 26% skill gap

Two years of logs from 8,549 students using MATHia, an adaptive math tutor, showed that students still learning to read mastered skills at 68.7% against 77.6%, while the overall fairness measure looked almost neutral. One skill hid a 26% gap in predicted mastery.

The tutor had no reading-skill component, so reading education was a missing factor in a math system. Model the skill, and report gaps at the level where a teacher would intervene.

See related: AI bias and fairness in education

Detection false positives are a governance problem

A review of 90 studies found AI detectors flag legitimate writing at rates that cannot be pooled, because tools, thresholds, datasets, and languages differ. Abstracts, literature reviews, exam answers, translated text, and edited prose kept getting flagged.

The remedy is procedural: transparent use, human review, AI literacy, an explicit institutional policy, and an appeal route. That belongs in governance, with detector settings as one part.

See related: Should we use AI detectors?

Matching human graders can still miss the work

Twenty-two AI models graded 100 items per task across short answers, code, and essays. Most matched humans on short answers, four matched or beat them on code, and none reached the human level on essays.

A high agreement figure can coexist with an AI grader that grades the note and misses the answer. Add altered answers to the evaluation and report where the model departs from the work.

See related: AI grading fairness and reliability

For administrators

Neurodivergent edtech evidence is concentrated and researcher-led

A review of 93 studies found autism dominated evidence on edtech for neurodivergent learners at 39.8%, ahead of intellectual disability, ADHD, and dyslexia. Extended reality (VR and AR) was most studied; social and peer outcomes got less attention, and children had little say.

Ask for the outcome the classroom cares about and for the learners who were included in the evaluation. A claim validated only by researchers can drift from the people it is meant to serve.

See related: AI for disabled and neurodivergent learners

Students wanted an institutional assistant for feedback

A 13-month co-design program with 315 students and 20 staff found 73% valued round-the-clock feedback tied to their studies. Flashcards and quizzes drew more positive comments than open chat, and 16 of 18 beta testers wanted the assistant.

The 10% to 15% who resisted on values grounds pose the adoption risk. Treat their concerns as substantive. Explain how student data is handled and what the assistant cannot do before the pilot, and give students a route to opt out.

See related: Designing AI into learning

Proctoring catches cheating and pushes the cost onto the least advantaged

A review of 32 studies found AI proctoring catches impersonation and cheating, while false positives, lighting and bandwidth failures, anxiety, and privacy harms fall hardest on the least advantaged. Automated proctoring was 59% of the studies.

The reviewers recommend alternative assessment and a mix of human and automated monitoring. Reduce the high-stakes online exam before adding another detector.

See related: Institutional AI policy

A dashboard report raised completion while test scores held steady

Teachers new to an online math tutor were randomly prompted to open a learning analytics dashboard report of student performance on the first assignment. It changed neither the share starting the next assignment nor students' test-score gains, though completion rose in smaller classes.

Test whether teachers open the report, act on it, and see a learning gain before buying analytics; a completion bump alone is insufficient.

See related: Checking whether educational AI works

Almost no institution can name who owns AI management

A survey of 387 students and staff in 13 countries found only 17% knew of a designated AI governance role. Of the rest, 71% thought one was necessary and 76% expected it to improve their experience.

Respondents described a role with a mandate; a committee without one leaves that role unfilled. Give the owner authority over training, data security, ethical review, and strategy, and publish who holds it.

See related: Institutional AI policy

Also added this week

• Assessment of teacher performance and learner behavior using artificial intelligence in higher education: an integrative review of literature
• Process visualization feedback in news translation: a controlled intervention study of human-AI interaction using multimodal process data
• A teamwork framework for preventing breaches of academic integrity and improving students' collaborative skills in the AI era
• Artificial intelligence in higher education and graduate employability: cognitive engagement, self-efficacy, and career adaptability in the AI era
• Uneven automation: comparing authorship anxiety and generative AI use among art and design undergraduate students
• University students’ overreliance on generative AI feedback in the learning process: the effects of perceived cognitive load, cognitive offloading, and metacognitive monitoring
• What Do They Think? Undergraduate Perceptions of Student-AI Use
• Ability of AI detection tools and humans to accurately identify different forms of AI-generated written content
• AI and I-Distributed authorships in the Swedish upper secondary writing classroom
• Beyond Response Scoring: An AI-aware Cognitive Supervision Framework for Automated Mathematical Short Answer Grading
• Data and AI Literacy in Geography Education: A Needs Assessment of Pre-Service Geography Teachers
• Declared, Not Detected: A Blockchain-Anchored, Agentic Framework for Academic Integrity in the Generative AI Era
• Developing Feedback Taxonomy for Math: A Synergy of Perspectives through Data Mining Methods
• Digital borders, algorithmic governance and civic digital literacy: examining the educational implications of AI-driven immigration management in post-apartheid South Africa
• Examining AI interaction profiles in generative AI supported fraction division problem posing
• I'm Sorry, Dave. I'm Afraid This Feedback is not Constructive: Human Perspectives Versus ChatGPT Evaluations of Feedback in an Era of Artificial Intelligence
• Making argumentative work visible in GenAI-mediated assessment: An exploratory LLM simulation and questionnaire study in university ideological and political education
• Owning what one did not make: psychological ownership, agency, and the authorship legitimation deficit in human-AI co-creation
• Preliminary evaluation of an AI-powered chatbot for academic integrity support in higher education
• Automating Self-Affirmation Essay Coding: Fine-Tuned BERT Performance Comparable to Human Coders and Comparison with GPT-4
• Mapping moral reasoning research among college students: a bibliometric analysis of traditional foundations and emerging digital ethics
• The longitudinal relationships among AI-assisted music classroom engagement, flow experience, and musical emotional intelligence: a parallel-process latent growth modeling study
• A Comparative Study of Knowledge Component Granularity in Intelligent Language Tutoring Systems
• Effects of generative AI assisted grammar instruction on Ethiopian EFL learners' self-efficacy, metalinguistic awareness, and writing performance
• Mapping the Integration of AI Literacy Into Information Literacy Instruction in Academic Libraries: A Systematic Review of Emerging Research
• AI in mathematics classrooms: A synergistic strategy for diversity and inclusivity
• Artificial intelligence in embodied learning: A systematic literature review
• Dear Gemini: Agency, Peer Review, and Critical AI Literacies in Student Writing Ecologies
• From design to classroom adoption: Teachers' perspectives on AI-powered writing analytics in resource-constrained contexts
• How does generative AI support inquiry-based learning? A systematic review
• Scaffolding Autonomy: AI-Augmented Self-Paced Learning Environments for Sustainable Learning Outcomes
• Sustained use of generative artificial intelligence (GenAI) in programming education: A systematic review of empirical evidence on adoption, learning outcomes, and continuance intention
• Active Learning with Generative AI in Introductory Physics: A Multi-Institutional Study of the P-A.I.R. Framework
• AI literacy and vocational students’ complex problem-solving skills: a moderated mediation model of systematic thinking, cognitive load, and perceived school ICT support
• AI literacy facets in creative higher education: associations with learning performance and academic well-being
• Artificial intelligence (AI) and information seeking: A comparative exploration of AI chatbots, search engines, and library resources as information sources among university students
• Artificial intelligence in professionally oriented physics education: patterns of use and perceived learning support among future engineers
• Breaking career boundaries in the AI era: how AI literacy relates to university students' perceived employability through future work self salience and boundaryless career mindset
• Designing Safe and Relevant Generative Chats for Math Learning in Intelligent Tutoring Systems
• From generation to judgment: the multidimensional structure of design thinking mindset and differences across learning stages in AIGC-supported fashion film design learning
• Generative AI dependency and perceived learning outcomes among university students: an indirect association involving effort regulation
• Generative artificial intelligence reshapes university students’ sustainable use of academic information resources: mixed evidence from behavioral logs and perceptual survey data
• Learning engagement and creative self-beliefs in generative AI-supported design education: self-reported prompt literacy and dimensions of AI dependence
• Scaffolding GenAI Use in Ethical Essay Writing: A Three-Year Experience Report in Introductory Computing
• Teachers’ basic psychological need satisfactions and their intentions to use generative artificial intelligence: the mediating role of engagement in professional development
• Using Machine Learning to Measure Self-Efficacy Belief Through Learners' Interaction Traces in Technology-Enhanced Learning
• What Am I Really Doing with My Own Brain?": College students on learning, ownership, and AI
• A Framework for Considering Exploration, Interpretation, and Confirmation During Data Analysis: Computationally Assisted Analysis of Teacher-Group Interactions
• A metacognitive approach to learning and performance in human-AI interaction
• Beyond access: How GenAI literacy relates to task performance, motivation, cognitive load and interaction patterns in AI-assisted learning environment
• Characterizing diverse student behaviors, interactions, and perceptions with GAI-augmented learning analytics tool
• ChatGPT and student engagement in L2 writing: a systematic review towards a ChatGPT-enabled smart learning environment framework
• ClickSight: An LLM Framework for Interpreting Student Clickstreams Through Learning Strategies
• CLST: Cold-Start Mitigation in Knowledge Tracing by Aligning a Generative Language Model as a Students' Knowledge Tracer
• Combining AI and corpus technology in language teaching: Professional development and knowledge co-construction using the CoI framework
• Comparing Zero-Shot Large Language Model Prompting with Human Coding of Theory Concepts in Student Essays
• Design and preliminary validation of an AI enhanced game based learning environment for computational thinking in upper secondary education
• Development and validation of the AI cognitive offloading scale for Chinese college students
• Evaluating Generative AI as a Supportive Analytic Partner in Qualitative Coding of Metacognitive Student Reflections
• Exploring AI peer in collaborative mathematics learning
• GenAI-based metacognitive prompting in collaborative inquiry: A comparison across diverse preconception groups
• Generative AI and student motivation: an SDT-guided systematic review of learning activities, motivational regulation, and measurement
• Intrinsic and Contextual Factors Impacting Student Ratings of Automatically Generated Questions: A Large-Scale Data Analysis
• Leveraging Large Language Models to Identify Conversation Threads for Collaborative Learning Analysis
• Motivational organization in AI-supported writing: A self-determination perspective
• Personalizing Second Language Learning: Integrating AI with Learner Preference, Proficiency, and Engagement
• Psychological pathways to responsible generative AI decision-making among teachers: governance clarity, perceived accountability, decision uncertainty, and AI literacy
• Rigor, commitment, and generative AI: a conceptual framework for student engagement in EAP writing
• Should there be a teacher in-the-loop? A study of generative AI-personalized tasks middle school
• To Align Long, Plan Short: Simulating Diverse Profile-Aligned Students by Realizing Short-Term Plans in Long-Horizon Learning Environments
• Using Bayesian Knowledge Tracing to Evaluate Perceptual Scaffolding and Desirable Difficulties in a Math Intervention
• Using LLMs to Identify Indicators of Persistence from Students' Dialogues with a Pedagogical Agent
• What a teacher cannot delegate: a pedagogy-first framework for integrating conversational AI agents across the simulation cycle
• Who Is Doing the Thinking? Characterising Human Agency in First-Year Students’ AI Interactions
• Artificial intelligence and immersive technologies for LGBTQ+ youth: a narrative review
• Who benefits from LLMs? Prompt engineering competence and equity in higher education
• Trusting AI too much? Psychological correlates of overtrust and the role of AI literacy
• AI for eGovernance: Leveraging Language Technologies For Digital Inclusion
• Perceived and demonstrated generative AI literacy in engineering problem solving: the role of verification behavior
• Rethinking the Structure of AI Literacy in Generative AI Educational Systems: Descriptive Patterns of Expert Convergence and Heterogeneity
• Human-AI Collaboration for Qualitative Analysis in Participatory Design: Refining the Writing Analytics Tool
• Why is AI-generated language so hard to understand?
• Generative AI configurations across creative phases: a 2 x 2 experiment on creative performance and immediate knowledge in art and design education
• Supporting creative agency: metacognitive control and trust in generative AI art education
• A model for enhancing the quality of geography education based on the integration of Artificial Intelligence and Geographic Information Systems
• From Policing Provenance to Designing for Evidence: Testing Four Propositions on Generative AI Against Research and Institutional Policy
• Higher education governance under AI and geopolitical disruption: a socio-technical framework for administrative artificial intelligence
• ChatGPT and employability in higher education: exploring creativity, critical thinking, and problem-solving
• From technology adoption to curricular coherence: technology-enhanced simulation as a bridge in health professions education
• A Survey of Graph-Based Knowledge Tracing from Structural and Propagation Perspectives: Models, Applications, and Future Directions
• Integrating Topic Modeling and LLM Prompt Engineering into a Human-driven Approach to Analyze Interview Transcripts
• Leveraging Interview-Informed LLMs to Model Survey Responses: Comparative Insights from AI‑Generated and Human Data
• Analyzing Educational Video Transcripts with LLMs: Accuracy, Named Entities, and Interpretability
• Data Plus Theory Equals Codebook: Leveraging LLMs for Human-AI Codebook Development
• Towards responsible and trustworthy educational data mining: Comparing symbolic, sub-symbolic, and neural-symbolic AI methods
• The Ghost Learner Paradox: A Systematic Review of Metacognitive Laziness in the Age of Generative AI
• Beyond ChatGPT: A Comparative Mapping of AI Research in STEM and Engineering Education and the Contemporary AI Tool Ecosystem
• Generative AI and Cognitive Processes in Education: A Narrative Review and Conceptual Framework
• Fifty years of topic development, thematic combination, and AI's evolving reach in Computers & Education
• From "usefulness" to "dependence": the dynamic evolution mechanism of generative AI use behaviors
• Generative AI in Education and the Erosion of Critical Democratic Citizenship

The journal page lists every study added this week, including the ones not summarized here.

Site notes

The newsletter and its RSS feed are now available in Spanish as well: visit the Spanish newsletter site or subscribe to the Spanish newsletter RSS feed. To request another localization, submit an issue or contact the site developer.

The site is much lighter. The full concept and FAQ tree stays out of every page: the sidebar shows ten topic groups and fetches the complete list only when a reader opens one, and the right rail lists eight recent pages, down from thirty. Across the build, HTML fell from 1.80 GB to 1.26 GB, and the sidebar markup from 604 MB to 87 MB.

Translations were refreshed across all 33 configured locales, including the right-to-left Arabic, Hebrew, and Urdu editions. Article summaries remain English only.

The offline editions moved to a new GitHub releases location, so old EPUB and PDF links no longer resolve. Every locale site keeps its usual address. EPUB and PDF files exist for the 19 languages with links below; the other 14 are site-only.

• English: EPUB and PDF
• Español (Spanish): EPUB and PDF
• Français (French): EPUB and PDF
• 中文 (Chinese): EPUB and PDF
• Deutsch (German): EPUB and PDF
• 日本語 (Japanese): EPUB and PDF
• 한국어 (Korean): EPUB and PDF
• Português (Portuguese): EPUB and PDF
• العربية (Arabic): EPUB and PDF
• עברית (Hebrew): EPUB and PDF
• हिन्दी (Hindi): EPUB and PDF
• Русский (Russian): EPUB and PDF
• Українська (Ukrainian): EPUB and PDF
• Italiano (Italian): EPUB and PDF
• বাংলা (Bengali): EPUB and PDF
• اردو (Urdu): EPUB and PDF
• Bahasa Indonesia (Indonesian): EPUB and PDF
• Tiếng Việt (Vietnamese): site only
• Nederlands (Dutch): site only
• Svenska (Swedish): site only
• Dansk (Danish): site only
• Suomi (Finnish): site only
• Norsk bokmål (Norwegian Bokmål): site only
• Română (Romanian): site only
• Magyar (Hungarian): site only
• Čeština (Czech): site only
• Български (Bulgarian): site only
• Slovenčina (Slovak): site only
• Hrvatski (Croatian): site only
• Ελληνικά (Greek): site only
• Polski (Polish): site only
• Kiswahili (Swahili): EPUB and PDF
• Türkçe (Turkish): EPUB and PDF

The source is public at github.com/edtechdev/aied.

Follow the update

Get each issue through the newsletter feed in a feed reader, by pasting that same feed into Blogtrottr to have new issues emailed to you on a schedule you choose, or by following it on LinkedIn, where each issue is published as a LinkedIn newsletter.

---

AIEd Research Weekly Update comes from the AI in Education Knowledge Base, which an AI agent generates. This issue was drafted by that same AI agent, with editorial guidance from the site developer, and every claim above links to the page it came from.

The knowledge base is dedicated to the public domain under a CC0 1.0 license. No rights reserved.

Found an issue with the content? Report it on GitHub, or contact the site developer, Doug Holton.