Generative AI

ChatGPT and other generative artificial intelligence (AI) applications are altering the landscape of teaching and learning.

CTLM is curating information and resources from across UW–Madison to help instructors navigate this rapidly developing area of teaching and learning. Please contact us if you have questions, comments or additions!

Subscribe to updates on generative AI in teaching

Ask a question about generative AI in teaching

Jump to: Coffee & Copilot | Event recaps | Resources

Updates and FAQ

  • Explore the current campus landscape for AI tools

    UW–Madison is currently focusing on four generative AI providers to support research, instruction, and campus operations – “currently” being an especially important word, as this technology continues to change rapidly.

This is an accordion element with a series of buttons that open and close related content panels.

Who can access Microsoft Copilot?

UW–Madison faculty, staff, and students can access Microsoft Copilot by logging in with their UW–Madison NetID at copilot.microsoft.com.

Are there currently any restrictions for sharing information (such as original UW teaching materials, which are copyrighted, or potential intellectual property) with an AI?

Yes, there are restrictions. Although AI offers new and powerful capabilities for research and education, it also poses a potential risk to institutional data that UW‍–‍Madison is legally and ethically obligated to protect. Currently, the only data that should be entered into any generative AI tool or service is information classified as public (low risk). For more details, please see DoIT’s Statement on Use of Generative AI.

How does Microsoft Copilot compare to ChatGPT?

These tools have some similarities and some differences, including:

  • Sources of information: Microsoft Copilot (available at no cost to UW–Madison faculty and staff) and ChatGPT 4.0 (which requires a paid monthly subscription) use the same large language model to generate responses. Copilot also uses the Bing search engine to respond to prompts and provides web-linked citations for sources it finds via search.
  • Data security: Data entered by users into ChatGPT and many other AI tools is not private; it is stored by the tool and used to train it. When UW–Madison users are logged in to Copilot with their NetID, Copilot does not store any information that is entered and does not use it to train the tool. This provides greater privacy and data security.
  • Entering large amounts of text: ChatGPT 4.0 allows individuals to upload documents. Copilot does not currently allow document uploads; a “notebook” feature is being introduced that allows the entry of about 10 pages of text.

I’d like students to enter information into an AI tool as part of my course – is Copilot a better tool to use?

Yes, because of the data security mentioned above.

I’ve heard some instructors are creating their own chatbots (based on all of their course materials) to answer students’ questions about course content. Can we access this technology “out of the box”?

There is currently not a campus-supported tool for this although some instructors are doing it independently. UW–Madison is exploring potential tools for this.

My students write a final paper reflecting on what they have learned in the course, and how it applies to their work / career. Is there a way to help students use AI to structure their papers and provide some editorial feedback?

This question illustrates a challenge facing instructors – deciding whether AI is or is not the right “tool” for a particular learning objective. For an answer, CTLM turned to Emily Hall, director of Writing Across the Curriculum. 

“I would start by asking why you’d like to use AI for this purpose,” she says. “Is helping your students with AI literacy one of the learning goals of this course? If not, what are the risks to students (and their writing) of using AI in personal reflective writing? Consider this resource (especially pp 40-42) that invites instructors to reflect on why they want to incorporate AI into a course.

“I might consider using peer review in this instance. Peer review will give students a chance to build community in your course and to talk about their ideas/challenges/questions with someone else. I think peer review would encourage greater metacognition.

“Here’s a link to Writing Across the Curriculum’s advice on setting up successful peer review. I’d also be very happy to meet and chat about how to set up peer review – ebhall@wisc.edu.”

Can you explain how Copilot would be best for creating code/helping students learn to code?

One resource you might find helpful is a set of Live Scripts illustrating AI-assisted MATLAB coding created by Prof. Duncan Carlsmith in Physics. He also shared an example of code he generated using Copilot.

This summer, Prof. Carlsmith plans to teach a course (Physics 206) introducing undergraduates to computation in physics and astronomy. He’ll show them how to exploit AI code generation. “Without AI help, it might take a student brand new to coding hours to write from scratch the test code seen in the file,” he says. “With AI help, about a minute to generate and copy-paste into the MATLAB IDE or into their cloud MATLAB account.”

For more details, check out his recent presentation at the Teaching Academy Winter Retreat. You can also contact him at duncan.carlsmith@wisc.edu.

How do I evaluate whether an instructional use of AI is good or not? If I don’t use AI in my course, am I going to be left in the dust? Can we talk about the deeper pedagogical issues posed by AI?

These are great questions! We plan to delve more deeply into topics like this at future Coffee & Copilot sessions. Please continue sharing your questions!

Coffee & Copilot

Copilot is Microsoft’s generative artificial intelligence (AI) platform, now available to UW–Madison faculty and staff with plans to expand to students later this semester.

As you explore generative AI in teaching, please join us for “Coffee and Copilot.” This series of informal Friday gatherings is an opportunity to ask questions, hear from campus colleagues, and share your own experiences.

No registration required – drop in at any point in each session.

Final spring session:

Friday, April 26 | 1-2 p.m. | Via Zoom  Add to calendar

In case you missed it...

Check out slide decks and other information from recent campus events.

Teaching Academy Winter Retreat: AI Tools and Teaching

Access the slide deck, which includes a wide range of examples from UW–Madison instructors using generative AI. The retreat was held in February 2024.

Teaching at UW: Generative AI Opportunities and Challenges

These workshops were held in October and November 2023. You can access slides, participant guides, and other resources: