David Nelson

A Theory of Adaptive Scaffolding for LLM-Based Pedagogical Agents Large language models (LLMs) present new opportunities for creating pedagogical agents that engage in meaningful dialogue to support student learning. However, the current use of LLM systems like Chat...

Article

arxiv.org/abs/2508.01503

I love this work. Biggest takeaways - AI Agents "often fail at effectively guiding students toward mastery" and "students prioritize scores over feedback, leading to off-task behavior that can hinder growth." Authenticity and relatedness are needed for these efforts. More humans in the process.

As models become tuned to specialized academic content, the gap between novice and expert ability to critically evaluate outputs will grow. This paper demonstrates the change in student error recognition from GPT4 to o3. More longitudinal studies like this please. arxiv.org/abs/2507.20995

Yep. The learning is definitely happening too slowly. That’s what little kids need for brain development. Speed. Jesus Christ these people.

LLM Agents for Education: Advances and Applications Large Language Model (LLM) agents have demonstrated remarkable capabilities in automating tasks and driving innovation across diverse educational applications. In this survey, we provide a systematic ...

It was relatively easy to keep up with the frontier AI models and a few open source clones. Agents are like electrical appliance manufacturers in the early twentieth century. What is actually useful? What is a niche product?
arxiv.org/abs/2503.11733

I lived it - in the 7th week I realized it, begged and was granted a gracious W from the professor - but that was a very scary phone call

Why can't we just have nice things! Why do our brains make us doubt this way?

Campus AI vs Commercial AI: A Late-Breaking Study on How LLM As-A-Service Customizations Shape Trust and Usage Patterns As the use of Large Language Models (LLMs) by students, lecturers and researchers becomes more prevalent, universities - like other organizations - are pressed to develop coherent AI strategies. LLMs ...

I've been focusing a lot on recently on trust and perceptions of trust + sycophancy + bot capability. Appreciate this preliminary approach for gauging Uni-owned bots vs Chat. arxiv.org/abs/2505.10490.

Increasingly convinced that simulation practice in clinical settings is one of the biggest "killer app" prospects for AI in education. Generate at scale, add nuance + complexity easily, personalize, etc.
iovs.arvojournals.org/article.aspx...

but that approach worked so well for ManU over the last half-decade......

Syllabus-bot Prompts Overview For an assignment in week 12, I asked students in our course to use the two bot profiles I had created for Nectir AI (Ted Lasso & Severus Snape). Students were to ask both bots questions abo...

Bot prompts and analysis of student feedback are here:
docs.google.com/document/d/1...

When Large Language Models contradict humans? Large Language Models' Sycophantic Behaviour Large Language Models have been demonstrating broadly satisfactory generative abilities for users, which seems to be due to the intensive use of human feedback that refines responses. Nevertheless, su...

I ran a small experiment with two syllabus chatbots - a Ted Lasso overly sycophantic version and a caustic Severus Snape version. Students really didn't like Snape's recalcitrance, but also hated Lasso's bland fawning style. We need more work on sycophancy and its effects
arxiv.org/abs/2311.09410

YouTube Chapter Generator - Generate summaries and timestamps for YouTube videos using Gemini API YouTube Chapter Generator - Generate summaries and timestamps for YouTube videos using Gemini API - chapters.py

gist.github.com/hamelsmu/520...

He shared at The Other Place, but I have to say here that @hamel.bsky.social 's YouTube chapter generator code is a godsend to academics who want to parse through hour-long video talks, especially when they concern AI and many are saying the same things. Link in comments

Large Language Models Don't Make Sense of Word Problems. A Scoping Review from a Mathematics Education Perspective The progress of Large Language Models (LLMs) like ChatGPT raises the question of how they can be integrated into education. One hope is that they can support mathematics learning, including word-probl...

This is solid math education work. "LLMs have mastered a superficial solution process but do not make sense of word problems." That superficiality and illusion of explanatory depth is one of the biggest dangers to student users who might think 'seems legit'.

arxiv.org/abs/2506.24006

A Brief History of AI: How to Prevent Another Winter (A Critical Review) Artificial intelligence has witnessed exponential growth in the past decade. Advances in computing power and the design of sophisticated artificial intelligence algorithms have enabled computers to ou...

www.pet.theclinics.com/article/S155... - Direct Link

My students appreciated Toosi et al's summary and framing within the context of AI Winters
doi.org/10.1016/j.cp...

It's good! Decent analogy (and one that is common to many discussions of deeper influences and causes).

"Check My Work?": Measuring Sycophancy in a Simulated Educational Context This study examines how user-provided suggestions affect Large Language Models (LLMs) in a simulated educational context, where sycophancy poses significant risks. Testing five different LLMs from the...

We've got decent work on sycophancy. This paper gets at the heart of the student use peril. Overconfidence in the tool AND yourself will redirect the output towards your conclusion. Makes it easier to dismiss the friction of error, and easier to outsource your thinking work. arxiv.org/abs/2506.10297

Purdue professors in the late 60s were apparently either Mary Poppins or Sam Kinison

Your Brain on ChatGPT: Accumulation of Cognitive Debt when Using an AI Assistant for Essay Writing Task This study explores the neural and behavioral consequences of LLM-assisted essay writing. Participants were divided into three groups: LLM, Search Engine, and Brain-only (no tools). Each completed thr...

This has been poked at in lots of places, so I won't, but to me it demonstrates the pitfalls of AI educational research when computer scientists omit learning scientists from their work, and vice versa.
arxiv.org/abs/2506.08872

information gain A place to gain a dose of AI information.

AI education is way too focused on LLMs and DNNs these days, excluding a lot useful topics in AI that are non-ML (e.g. search, planning, knowledge, etc.). That's one of the motivations behind this series I started, www.informationga.in/blog/ai-expl..., which is a first principles treatment of AI.

The league is going to allow them a extra midfielder who doesnt count among the 11

The latest open artifacts (#10): More permissive licenses, everything as a reasoner, and from artifacts to agents Artifacts Log 10.

Beyond LLM metrics or any gauge of effectiveness, the faculty I have worked with vibe first with the voice, syntax and prosody trends of each bot.

DeepSeek spurred smiles and curiosity among many colleagues.

Big hat tip to: @natolambert.bsky.social

www.interconnects.ai/p/latest-ope...

I argue for a broader approach here:
academyforeducationalstudies.org/wp-content/u...

Generative AI in Graduate Bioprocess Engineering Exams: Is Attention All Students Need? State-of-the-art large language models (LLMs) can now answer conceptual textbook questions with near-perfect accuracy and perform complex equation der…

Would like more studies like this that quantify the impact of active AI use on exams. While this is only part of the puzzle, I think, modifying learning outcomes beyond disciplinary recall benefits from detailed study of how gameable current assessments are. www.sciencedirect.com/science/arti...

I wish they would introduce a mode a la Google Incognito - where it won't take any previous queries into account, thought it will store your prompts and the bot's responses

Hush was my immediate thought - think it holds up well. But I like dark willow saga most

Which season is pulling the most?