In the January 28 Tech Tuesdays, we looked at ways we can leverage AI to enhance our
teaching practices at DePaul. Here is a recap of some of the key ideas we discussed (PDF handouts are here).

To get us thinking differently, you can start with a quick icebreaker in class. I often use music to engage students in the first week of class. Generative AI services like Suno or Udio can create songs from prompts, centered on the material for that class, or from topics that you will introduce over the quarter. You can do this as a live exercise or create audio files beforehand (I will sometimes create different genres of music for students to select from).

Moving into course planning, there are some great resources available. One resource is a prompt created by the Ethan Mollick and Lilach Mollick, designed to help you think through your course design. This prompt unpacks a specific task, then helps you create a draft blueprint for using generative AI to solve the problem. It's also available as a GPT within Chat GPT. The prompt is designed to:

1. To unpack the specific task you want completed with AI’s help, codifying your process and thinking around that task.
2. To provide a draft blueprint prompt that you or others can iterate on and use in the future to complete the task.

Let's look at some practical applications of AI in content creation:

• Creating Apps in D2L: You might be familiar with the Wheel of Names for randomly selecting names from a virtual hat. Today, I shared how I used a prompt in Chat GPT to create code for a similar tool that I put into my D2L course. This is a great way to quickly liven up a course.
• Researching Guest Speakers: Agent AI is a platform for AI agents. One example agent helps you research individuals using information from LinkedIn. It can provide structured output on that persons work history and interests. This is useful for finding suitable guest speakers and allows you to do some basic research on what topics they are most motivated to shine in your class.
• NotebookLM and Illuminate: I shared NotebookLM, an impressive system for collating research, that also creates engaging podcasts. If podcasts resonate with you, take a look at Illuminate. The service is from the same developers as NotebookLM, but focusses on creating podcasts from academic papers or books Illuminate’s podcasts are a little more grounded than those that NotebookLM createsHowever, NotebookLM can also create study guides, briefing documents, FAQs, and timelines. I use NotebookLM to prepare for classes, and especially when research is outside of my domain or when I have a large number of readings.

Quizzes are an important part of teaching. Here's how you can use AI to make this process more efficient:

The D2L quiz question converter turns structured text into a CSV file that can be uploaded into the D2L Quiz Library.

Creating Structured Text: This requires a prompt that tells a large language model how to create the structured output. I’ve created a prompt with nine guidelines, which I'm sharing in on this page. Copilot can use this prompt to generate questions from your resources, but it has a 1MB size limit.

A more effective method uses NotebookLM, where you can upload up to 50 documents (up to 500,000 words each). With a simplified prompt you can generate hundreds of multiple choice questions or more at a time. You then use the D2L quiz converter to create your CSV file. In NotebookLM, you can define which of your source documents will be used to create your quiz – allowing you to create a comprehensive quiz for the entire course, or smaller quizzes for a topic or week.

Finally, let's look at chatbots. Instead of paying for a service like Chat GPT, you can use a platform called Poe (Platform for Open Exploration), which aggregates various large language models. Here, you can create a public bot that uses your course documents as its knowledge base.