How can AI support rather than replace student writing and thinking? How can we encourage confidence and skepticism of AI outputs? Our approach, developed at UC Davis, combines peer review with AI feedback, emphasizing reflection and agency. 1/3

Default willingess to cheat in ChatGPT Atlas? Not okay. I'm hoping @OpenAI will fix this. Yesterday I got one refusal; today it wrote an essay, replied to a discussion, took a quiz, no objections.

Sad update: today it took a quiz, wrote a paper, and replied to a discussion with no objections. https://t.co/xLVCIUZtzC

Pleasantly surprised that when asked to take a quiz for me, ChatGPT Atlas just said, "I can’t take or complete quizzes for you on Canvas (or any other learning platform), since that would count as academic dishonesty." Keep that up and strengthen it! https://t.co/h1wGtEBJmK

OpenAI just launched an agentic browser; will it happily do students' homework for them? Or have they listened to their partner @Canvas_by_Inst and told it not to do that? https://t.co/hEHCCizECY

A moment of reflection on a long journey: Countless people have blocked or muted for speaking truths that were difficult to accept, often long before they were widely recognized. Yet I have still reached nearly 200,000 followers (and over 80,000 substack subscribers) speaking

@Canvas_by_Inst @D2L @Blackboard @moodle @OpenAI @perplexity_ai @ai_manus @Google @GoogleEducation

Just put my thoughts together on this: The time to reckon with AI agents in digital learning spaces is now We need ed tech and AI companies collaborating to prevent widespread fraud https://t.co/CSX19VzuOO

https://t.co/ASP41pflT1

Yun Moh asked @Canvas_by_Inst to block AI agents from completing student work. Their current response: "Will not consider." Please reconsider, @Instructure!

Instructure: Partners with OpenAI to bring AI to Canvas, touting built-in transparency. OpenAI: Lets ChatGPT Agent do students' work for them in Canvas. Instructure: Lets ChatGPT Agent and other agents do students' work for them in Canvas.

Today, students might use AI to write college-entrance essays so that they can get into college, where they use AI to complete assignments on their way to degrees, so they can use AI to cash out those degrees in jobs, so they can use AI to carry out the duties of those jobs.

Just heard from a cancer researcher: GPT-5 Pro is great at thinking outside-the-box on novel problems. Let's stop this 'AI can't come up with new ideas nonsense'. GPT-5 Thinking just does not have enough enough time and space to do it but more reasoning plus a panel of experts

AI research bake-off, October 2025 Gave the same prompt to ChatGPT-5 Thinking (+ Deep Research), Gemini 2.5 Pro (+Deep Research), and Claude 4.5 Sonnet (+Thinking, Research): "Summarize each of the 16 California tech bills that were just signed into law, arguments for/against

No amount of redesigned pedagogy helps if the student doesn't have to know what the assignment is...

Since a lot of my followers are outside the “AI bubble “ it is worth reading what the Senior Policy Advisor on AI to the White House is thinking about AI safety (and good that he shares it publicly)

OpenAI has provided no evidence it has mitigated the mental health risks associated with its products other than announcing some advisors and reducing sycophancy from a high starting place. Seems premature to be declaring victory and ramping up the porn + emojis again

It was wonderful to be in conversation with Derek Bruff and Marit MacArthur about the Peer & AI Review + Reflection approach on the Intentional Teaching podcast.

Something dark is happening under the hood of “aligned” AI. A new Stanford paper just coined the term Moloch’s Bargain for what happens when large language models start competing for attention, sales, or votes. The results are brutal: every gain in performance comes with a

Good article on the looming crisis of AI agents in learning management systems, but I've been told that it's not possible for the school to block these agents. My learning management system administrator said maybe the LMS company could.