Jan. 29, 2024
How can students and instructors use generative artificial intelligence (AI) to create and assess content for writing assignments? This salient topic was discussed in “Generative AI and Teaching: Uses in Assessment of Learning,” a panel hosted by The Sara Fine Institute and the Research, Ethics, and Society Initiative on Jan. 19. The event's facilitator, Dr. Marcia Rapchak (teaching assistant professor, Department of Information Culture and Data Stewardship) was joined by panelists:
- Dr. Diane Litman, associate dean for mentoring and development, SCI; senior scientist, Learning Research and Development Center
- Dr. James P. Purdy, professor; director, University and Community Writing Center, Duquesne University
- Dr. Annette Vee, associate professor, Department of English; director, Composition Program
“Instructors should highlight the need for equity and transparency in assessment,” said Purdy. “It’s important for instructors to talk to students openly about what are permissible uses of generative AI in their courses and in what contexts.”
Rapchak asked the panelists how involved generative AI should be in academic settings and how students and instructors should use them. Litman compared generative AI to a Teaching Assistant (TA): like a TA, generative AI could be used to assist assessment, rather than as a substitution for instructor involvement.
“Generative AI needs instructor involvement and training. It’s very important that teachers audit its capabilities to get a sense of how well it’s actually working. That might relate to how much you can trust it,” explained Litman.
Systemic bias is an ethical implication of using generative AI. As with many other AI algorithms, biases within the system may exist that could harm students based on race, class, or gender. Rapchak asked the panelists how these ethical implications could be mitigated.
“I don’t think we can train bias out of these systems. I think that we can work with them to use generative AI more appropriately in different situations,” said Vee. “These systems should be transparent and held accountable so that it’s not just a black box proprietary algorithm making decisions.”
The event concluded with discussions of future implementations of generative AI.
“I think considering opt-in and opt-out models for use of these generative AI technologies can be helpful,” said Purdy. “I think it’s problematic if we use these technologies without letting students know that we’re using them. Students need to have some agency if they’re not comfortable with it and have alternative mechanisms in place.”
Watch the video of the panel here.
--Alyssa Morales