Skip to main content


My unscientific #AI test question:
How can #teaching be tailored to student learning styles?
Only Claude gives a correct, critical answer:
* Gemini g.co/gemini/share/10d121293154
* OpenScholar openscholar.allen.ai/query/315…
* ChatGPT chatgpt.com/share/e/6741bb7b-0…
* Claude: "The concept of distinct "learning styles" (e.g., visual, auditory, kinesthetic) actually isn't well-supported by research evidence, despite its widespread popularity in education..."
#AIEd #EdTech
in reply to Doug Holton

I asked Chat GPT two questions. About the neuromyth learning styles. weblearning.co.za/blog/debatin…
in reply to Doug Holton

what do you think of the result if I tweak the prompt just a touch,

g.co/gemini/share/24b977072026

One of the most effective ways of writing prompts involves telling the LLM to answer from the point of view of an expert and not the Internet as a whole.

It doesn't wholly correct the error, but I think it's a "When did you stop shoplifting?": style of question.

in reply to Stephen Cerruti

@scerruti that's a good point, I guess I was surprised though that even OpenScholar, which more rigorously searches for relevant research, uncritically argued for learning styles. In the system prompt for my college teaching coach bot on huggingface, I specifically ask it to cite its arguments, but when I tested it on this question, it came up empty, saying it didn't know. It searches the web in general though, not just research.