Analyzing an AI-Generated Podcast

Overview

Students compare an AI-generated podcast with the source document used to create it.

Why Use This?

This activity requires students to critically analyze content generated by AI while also giving them an opportunity to engage with the source material more deeply by verifying it has been correctly summarized. Importantly, the activity also reinforces the idea that we shouldn’t inherently trust AI-generated content and that it requires careful review.

How Does It Work?

Students are presented with a source document and a podcast that has been generated from it using Google NotebookLM’s Podcast Generation feature. They are then asked to listen to the podcast and determine how accurately it has captured the information in the source document. Finally, they present their findings and reflect on their experience in a discussion.

Keep In Mind

Make sure the source document is manageable in terms of length and complexity. Research articles are great to use, but other complex academic content works as well. Avoid using blog posts and other short-form articles that are not detailed enough to make the activity valuable.
Start with documents that are 3-10 pages in length before moving on to longer materials.

Testimonial

I think this activity provides an interesting way to scaffold more complex source materials. As instructional designers, we often suggest that faculty annotate their resources; however, this activity takes annotations to the next level by providing a summary of a complex resources in podcast form and then having students analyze/verify the output. We get 1) scaffolding and 2) spaced repetition.

Nate Ewings, Instructional Designer