AI and Large Language Models: What do they mean for the future of teaching and learning? by Katie Hayes

About

katiehay — 2025-09-05 13:23:09 UTC

Large Language Models (LLMs) such as ChatGPT and Claude have quickly burst on our scene. They are based on very large bodies of online text, but they also reflect the biases and limits of that online world, where white male Americans interested in technology are overrepresented.

LLMs raise fundamental questions about the nature and future of education, such as: What does it mean to know something? What are the implications of these tools for how we learn, how we teach, and how schooling is organized? What can other technology-driven educational reforms tell us about the likely consequences of LLMs? What human skills will be valuable in a world where LLMs exist, and how can we ensure that we and our students acquire them? Each of these topics raise challenges for our mission of promoting equity in learning and teaching.

Each panelist will give brief presentations on these questions and then our community will have the opportunity to discuss these topics and concerns about AI and Large Language Models in education.

I am concerned, based on some preliminary studies (Kosmyna et al., 2025; Stadler et al., 2024), that significant usage of LLM-based chatbots will lower students’ cognitive load and thus decrease their learning and cognitive abilities (e.g., memory, critical thinking, etc.). Do you believe this is a reasonable fear, and if so, what will the long-term effects of LLM-based chatbot usage be on cognition?

2025-09-05 19:29:24 UTC

dkang33 — 2025-09-05 20:20:50 UTC

There are a lot of questions on genAI, LLMs and the impact to the "now" which honestly educators are late to the game and will take time to figure out. However with multimodal around the corner, what are your thoughts on preparing for the next phase?

Catching ‘Statamon’: Gamifying Stata Learning

2025-09-06 23:46:55 UTC

An example from my personal use:

I’m using AI to gamify my learning in how to code for Stata, the statistical software.

My approach is inspired by the video game Pokémon: in the game, players search for creatures, add them to an encyclopedia (Pokédex), and attempt to capture them. In my version, I seek out “Statamon” to add to my personal “Statadex”.

When I ask the AI what a command means or how to code a particular action, it generates an encounter. That encounter provides me with a Statadex entry (which I track in Notion). The entry is a mix of technical explanations with Pokémon inspired flavor text. Over time I hope to build out a personal coding reference system that’s actually enjoyable to refer back to.

I also use the AI to generate coding challenges or quizzes, which is used to catch or train the Statamon. It’s still a work in progress, but so far it’s been a surprisingly fun and motivating way to deepen my Stata skills

2025-09-09 17:38:56 UTC

Where would you send educators and parents for guidance on the appropriate use (or not) of LLMs by young people for learning?

Which organizations produce the best content to inform general audiences on these topics?

K-12 Lesson Planning Analysis

karenlynnyoung — 2025-09-11 01:30:43 UTC

When working with educators, I use LLMs as collaborative tools for developing, generating, and analyzing lesson plans. I have them write a lesson plan on their own. Next, we construct a prompt asking for a lesson plan to be developed on the same standard. We experiment with providing the system detailed information about the unique characteristics of their classrooms and compare the lesson plans to the one they have written. We carefully analyze the generated lesson plans for accuracy, potential bias, overall quality, etc. Teachers then reflect on and examine the suggested activities and consider modifications. We have deep discussions about the implications of having LLMs create a lesson plan and then teaching the lesson immediately without taking the time to analyze what was produced.