Investigating Teachers' Views of Potential Bias of Artificial Intelligence After a Professional Development Program (Poster 7)

• In Event: Opportunities and Challenges in Teaching Algorithmic Justice in K–12 Computing Education

Abstract

Objectives: We present an innovative professional development (PD) model, called Everyday AI (EdAI), designed to advance teacher AI literacy, which includes an understanding of the ethical issues posed by bias in AI as foundational knowledge. Here we describe the design of the EdAI PD and findings on teacher experiences, including their thoughts on teaching ethics in AI and their motivation for teaching AI in their classrooms.
Theoretical framework: The design of EdAI PD was influenced by the Costanza-Chock (2020) design justice framework, which calls upon designers to work closely with community organizations and the communities they serve to challenge structural inequalities. Through a series of ten weekly online meetings in the style of a book club and asynchronous discussion of the book Artificial Intelligence: A Guide for Thinking Humans by Melanie Mitchell, teachers fostered a community that encouraged its members to think critically of AI and to ask questions about bias as they explored lessons and activities from the Developing AI Literacy (DAILy) curriculum together.
Methods & Data Sources: We interviewed 24 educators about their experience of the PD. The interview questions probed teachers’ experiences (challenges, successes, and suggestions) of each of the PD. With participant consent, we also gathered conversation data from the Slack channels teachers used for asynchronous communication during the book club. Each week, a team member volunteered to pose a discussion question to which the group responded. In total participants created 260 discussion posts on the Slack channels. Teacher interviews and Slack discussions were analyzed separately using thematic analysis to identify trends in teachers’ experiences of the PD.
Results: Analysis of teacher interviews and Slack discussions revealed unanimous agreement with the framing of AI as a social justice issue and that AI education should emphasize how AI is impacting everyday life. For example, a female teacher posted in Slack that she thought AI bias can impact her and her community, “As an African American woman, the AI bias regarding people of color and women is extremely alarming and I am cautious of trusting the incorporation of AI in any known area of my personal life.” As another example, an educator shared in the interview that AI ethics is “important because everybody should understand some of the algorithms weren't meant to be biased, but they are inherently biased because of the data sets that are used. People need to be aware that because of those limitations of the data sets, the decisions that are output based on that algorithm could impact them or people that they know.”
Significance: In EdAI, teachers from historically marginalized groups as well as teachers of students from historically marginalized groups encountered the concept of bias in AI during the EdAI PD and came to see it as one of the social justice issues of our time. By the end of the PD, participating teachers universally expressed a commitment to teaching AI as a means of raising community awareness of the potential harm of bias in AI to their students and families.

Authors

• Irene A. Lee, Massachusetts Institute of Technology

• Helen Zhang, Boston College

• Katherine Strong Moore, Massachusetts Institute of Technology