Developed By Jennifer Gradecki

Course Subject:	ARTH 1100: Interactive Media and Society
Student Level:	Undergraduate
Number of Students:	Up to 17
Developed by:	Jennifer Gradecki, Associate Professor, College of Art, Media, and Design

What Students Did

In this workshop, students learned about the negative impacts of AI-generated disinformation campaigns. They learned through a student-led article discussion, a lecture on the topic, and through reflecting on how they might recognize and respond to AI-generated content they might encounter in the future.

Purpose

The workshop served dual purposes. From a media and society perspective, the activities were designed to help students think critically about the risks that AI generated content–including text, images, audio, video, and deepfakes–can contribute to disinformation, polarization, and truth decay. From an AI literacy perspective, students learned how such content is created and how they can recognize it.

Assessment

The discussion portion of the workshop was done verbally, and the reflection portion of the module was done in writing, through a discussion in Canvas. The assessment of both components was focused on how well students understood the core module content, including how large language models (LLMs) can be used to create disinformation, how content created by LLMs might be received and disseminated on social media, and how students might engage with this kind of content if encountered.

Faculty Reflections

Students found it difficult to discern whether some content was generated by an LLM or by a human and benefitted from the discussion of ways to discern LLM-generated content on social media. They found AI-generated text to be the most difficult to identify.

In their written reflections, some skipped or glossed over the most challenging questions, which pertained to emerging technologies for image generation that make the output more similar to those created by humans. For this reason, I revised the reflection assignments to require responses to these questions. This resulted in more robust responses.

This technology-specific workshop will require frequent updates to keep pace with rapid developments.

Step-by-Step Directions

Step 1	Students read an assigned article about potential negative impacts of generative AI (e.g., disinformation campaigns).
Step 2	One student presents the article and facilitates a discussion about it. NOTE: In each module, a different student leads the discussion so that all students in the course have an opportunity to lead.
Step 3	The instructor follows up with a brief summary of the discussion, filling in any gaps in information and correcting any misconceptions.
Step 4	The instructor provides a mini-lecture on the workflow for AI-generated disinformation campaigns, including topics such as: Technical manipulations Strategic targeting of particular groups Potential social impacts Examples of projects that address the problem Example discussions of LLM-generated content. Sample prompts: If this content appeared in your social media feed, how might you react to it? Would you be able to tell that it was generated by AI? If so, how could you tell? If not, why not? Do you think you have encountered AI-generated disinformation in your own social media feed?
Step 5	Students respond to the following reflection prompt (which includes a link to class slides) in a Canvas discussion: In response to our in-class discussion on Generative AI and disinformation, please provide a 250-word or more written response to the following questions: What were your key takeaways from the in-class discussion about how generative AI can be used to create disinformation? If you came across AI-generated content like the examples we looked at and discussed in class in your own social media feeds, would you be able to tell that it was AI-generated? Please consider both text and images. If so, how? What techniques would you use? If not, why not?

Related Materials

Assigned Article: Helmus, Todd C., Bilva Chandra. “Generative Artificial Intelligence Threats to Information Integrity and Potential Policy Responses,” RAND Expert Insights, Published Apr 16, 2024.
Examples of public projects that address AI-generated disinformation campaigns:
- Generative Persuasion interactive artwork featuring an LLM-based app that produces micro-targeted disinformation: https://jennifergradecki.com/GenerativePersuasion.html
- Tactical Tech’s Data Detox Guide about AI-powered scams: https://datadetoxkit.org/en/ai/tricks/

Download PDF version of this assignment

Return to AI Gallery Home