Recap: Winter AI Institute for Teachers
Recap: Winter AI Institute for Teachers
Last week, CETL partnered with the Department of Writing and Rhetoric to offer a second iteration of the AI Institute for Teachers to an audience of UM instructors from across disciplines. Nearly 60 faculty from 26 different departments and schools attended the three-day event. In a wide variety of interactive sessions designed by Institute leader Marc Watkins, participants examined the impact of generative AI on teaching and learning, working in small groups to consider how to approach AI in their own disciplines.
If you’re not a UM faculty member or couldn’t attend the sessions, we have good news! All the materials from the Institute are publicly available at the following links:
And we’ve written a short recap of the Institute here.
Day 1: AI Literacy Basics
The first day of the Institute provided an introduction to the fundamentals of generative AI. Marc began with readings and activities designed to show participants how large language models (LLMs) are trained and prompted—including one activity that resembled playing Pictionary with AI! They then explored activities that asked users to write “with” and “against” AI text generators, work that might help students contemplate and comprehend this new technology.
The second session of the day explored the myriad ethical challenges presented by generative AI, from potential copyright violations to insidious surveillance or the spread of misinformation. Participants discovered their facility (or lack thereof) in identifying AI-generated images with the game “Which Face is Real?” and had lively discussions about AI ethics and “what-if” scenarios that involved fields like healthcare, law, and journalism.
During lunch, CETL director Josh Eyler and I led a conversation about why the rise of generative AI calls for new approaches to grading. We discussed how alternative grading can help discourage inappropriate use of AI and shared resources, like CETL’s alternative grading bibliography, that might provide interested instructors with new models for their own classes.
Marc closed the day with a session on AI detection, noting its unreliability and potential for bias. Participants tried their own hand at identifying human- and AI-generated text with the “Real or Fake” game and tested out a variety of AI detectors on their own. It turns out that identifying AI-generated text is harder than it seems.
Day 2: Applying AI Assistance
The second day of the Institute was devoted to applications of AI in education, examining use cases for AI writing, reading, research, speech recognition, and feedback. Participants began by experimenting with a variety of LLMs, using them to summarize texts, transform writing into different modalities, and find relevant research. They also honed their skills in prompting and even built their own chatbot personas.
In the following session, Marc and the group explored the AI writing assistant Lex and AI reading assistants like Explainpaper and SciSpace. Participants had a chance to speak with their colleagues, and with a variety of fun chatbot personas, about the implications of these programs for their courses and disciplines.
At lunch, I introduced another CETL resource: our syllabus template with suggested language for generative AI policies. This brief session asked participants to consider how to craft a policy that made clear to students what tools they might use, how they might use them, and (importantly) the reasoning behind these decisions.
Marc closed the day with a discussion of AI research assistants like Perplexity and speech recognition tools such as Otter.AI. Like the other tools discussed during the institute, these programs pose a number of potential ethical challenges and may create obstacles to learning—but they can also be transformative in supporting learning, especially for students with disabilities. The thorny dilemmas posed by the day’s activities led to rich discussions among participants.
Day 3: Advanced AI Frontiers
The final day of the AI Institute moved beyond ChatGPT—way beyond—to explore AI image, video, and music generation. Participants began the day by generating their own music with interactive tools like Blob Opera and MusicFX and their own videos with the platform Runway. They also discussed the far-reaching, sometimes hilarious, and often troubling implications of these technologies.
The second session covered the hot topic of using AI to “talk to your files,” a process sometimes called “Retrieval Augmented Generation (RAG).” The group uploaded sample files to Google’s NotebookLM and Anthropic’s Claude and asked the programs to run sentiment analyses, generate study guides, extract keywords, compare documents, and more.
In the final session of the Institute, participants considered the potential of specially-trained chatbots for educational use. Marc demoed a chatbot, designed by himself and his students, called the The Grove Squirrel: a loveable “personal tutoring assistant” for University of Mississippi students. The group ended the day by experimenting with Poe to train their own chatbots.
The AI Institute was a three-day whirlwind, and I mean that in the best possible way! Learning about recent advances in generative AI can be like drinking from a firehose, and I can’t help feeling that I and many other participants left the Institute with more questions than answers.
But that’s okay: generative AI poses a lot of difficult questions for our lives within and beyond the classroom. The AI Institute was a good reminder that there are no easy solutions to the teaching problems we face. And that helping our students, and ourselves, become better consumers of media, information, and technology is more important than ever.