"I Feel Like Police Aren't Qualified to Judge Algorithms": Artificial Intelligence Ethics Project-Based Learning in a Politically Divided Community

• In Event: Project-Based Learning of the Future: From Equity-Adjacent to Equity-Infused

Abstract

Purpose

This study examines an artificial intelligence ethics project conducted in two online 9th grade ELA classes in a community where recent attempts at social justice oriented instruction have been met with resistance. AI technologies tend to reproduce human bias (Buolamwini & Gebru, 2018) and have played a role in election interference (Kamark, 2018). These are topics that our team, including our partner teacher, deemed likely to be contentious in our target community. This study seeks to answer the question, “How can we implement justice oriented artificial intelligence ethics projects in potentially resistant communities in a way that sparks critical inquiry in students and protects teachers from undue criticism?”

Perspectives

We build upon our enhanced literacy PBL framework that values, leverages and extends all the languages and literacies that students bring to school as they use technology to explore and respond to social problems (Author, 2021). We also draw upon Lee and Soep’s notion of critical computational literacy (2016), which describes skills that counteract inequitable impacts of technology upon our society.

Methods & Data Sources

This project introduced students to AI, algorithms and ethics and explored AI ethics topics including algorithmic bias and privacy vs. security. We used original short stories, non-fiction texts and hands-on AI development experiences to build understanding of AI technologies and their impact on society. Students created digital artifacts to express their ethical stances and used those artifacts to teach AI ethics lessons to middle school students.

We sequenced learning experiences so that students had a strong understanding of underlying technology and had the chance to discuss less sensitive topics before moving onto more divisive ones, such as racist predictive algorithms used in the criminal justice system. This approach aimed to support students in developing and expressing critical perspectives on AI technologies, protect vulnerable students from harmful interactions, and reduce the likelihood that our partner teacher would face criticism from parents or school administrators.

Data sources included student work products, surveys, interviews, field notes, and chat transcripts. The study took place in a rural, predominantly White high school in the Mountain time zone.

Results

Students demonstrated increased understanding of AI systems and their potential to amplify or alleviate social problems. Students chose to create media focused on negative uses of AI, such as social media manipulation and AI replacing human workers, as well as positive examples, such as AI that helps doctors diagnose, or detects forest fires using satellite imagery. Supporting inquiry into a range of topics was vital for student investment. Our approach to sequencing AI ethics topics, coupled with strong and clear classroom norms established by our partner teacher, allowed us to explore politically sensitive topics without inducing conflict within or beyond the class.

Scholarly Significance

While research centering equity in project-based learning has increased in recent years, this study helps fill a gap concerning how to do this work in resistant spaces. Given efforts in some States to constrain what students are allowed to read and talk about, this is an urgent concern.

Authors

• Benjamin Walsh, University of Colorado - Boulder
  Presenting Author

• Bridget Dalton, University of Colorado - Boulder
  Presenting Author

• Stacey Forsyth, University of Colorado Boulder
  Presenting Author

• Tom Yeh, University of Colorado - Boulder
  Presenting Author