A Rhetorical Framework to Detect Bias and Misinformation in AI Outputs

Outline

Google Doc Link for easier reading: https://docs.google.com/document/d/13pE-taTCMekjkVjKnRq1EeaQ4C8TEhVtxMEbDerJnSI/edit?usp=sharing

Introductions (1 minute)
Learning Objective: Participants will learn rhetorical analysis skills to evaluate AI-generated content. Participants will Integrate these strategies into lessons to enhance media literacy and critical thinking. (1 minute)
Icebreaker: Use Padlet “I can’t draw” to generate an image of their favorite animal doing their favorite activity in their favorite vacation spot. (5 minutes)
Mini-Lecture on Bias and Information:
Define and discuss the concepts of bias and misinformation. (5 minutes)
Explain how AI outputs can perpetuate stereotypes, bias, assumptions, misinformation
Audience Participation: Provide an example of an AI generated text or image (provide a Choice Board) and ask participants to spot/identify how they can tell it’s AI. We will gamify this! (5-8 minutes)
Mini-Lecture on Rhetorical Analysis:
Define and discuss 3 rhetorical elements to consider when evaluating AI outputs for bias and misinformation. (3-4 minutes)
Audience
Purpose
Message
Activity: Participants will use either Gemini, Raina (MagicSchool), or ChatGPT to generate a response to an ethical question. (10 minutes)
Participants will volunteer to share the AI generated response in the chat and presenters will evaluate the response for at least one rhetorical element. (2 minutes)
After presenters model how to evaluate, participants will evaluate their own responses by answering the following questions: (5 minutes)
What is the purpose of the AI-generated content?
What messages are being conveyed, and how are they being framed?
Who is the intended audience for the content?
Mini-Lecture on Psychological Dangers of AI revising one’s writing
Writing Modes: Personal Statement, Narrative, Poetry, Journaling (1 min)
Vulnerability and lack of confidence in writing about yourself/feelings (2 min)
Current research / Quick review of theories via quotes from experts (2 min)
Example of AI’s revision suggestions for several 2-3 sentence narrative pieces of writing.
Highlight areas of assumption/stereotype, bias, etc (5-7 min)
Have participants go pack to the Padlet and generate an image by telling AI what they looked like in high school. Use as many details as possible.
Discuss results. Were they accurate? What did they include? What did they omit? What was emphasized? (3 min)
Collaborate with Audience: Discussion of the dangers to self-esteem as a result of the bias. (3 min)
Closing
Next steps (2 min)
What can you do to empower learners to better evaluate AI outputs?
How can an understanding of rhetoric develop the skills needed to evaluate and analyze AI outputs?
Presenters will provide several sample lesson plans for educators to take back to their classrooms.

Supporting research

Autonomy & Identity in the Digital Age (SagePub 2016)
https://www.sciencedirect.com/science/article/pii/S0190740916303280?casa_token=wYOtzZdcE7EAAAAA:lzslfDX3zi5QECPeoYldeF2tz2_J7cZX2WSChLTgW5Bq6R5XiMpP5yI4RO9QS2AMdMQKQI6Fzg

Adolescents’ wellbeing using AI powered ACT tool - longitudinal study (PsyarXIV, 2022)
https://psyarxiv.com/mpx3v/download?format=pdf

Developing Middle School Students’ AI Literacy (SIGCSE, 2021)
https://dl.acm.org/doi/abs/10.1145/3408877.3432513

Ethics for AI Writing: Importance of Rhetorical Context (2020)
https://dl.acm.org/doi/abs/10.1145/3375627.3375811?casa_token=TuoF7Y2VL1QAAAAA:zfolh3hgZVZUsM1iDJJ3cMVu3b4f_4_PL5Us7iOiXe8lR6rGG7nKoygSMgRnJbcXQmANCnxLKIf8