A critical appraisal of equity in conversational AI:
Evidence from auditing GPT-3’s dialogues with different publics on climate change and Black Lives MatterEvidence from auditing GPT-3’s dialogues with different publics on climate change and Black Lives Matter
Autoregressive language models, which use deep learning to produce human-like texts, have become increasingly widespread. Such models are powering popular virtual assistants in areas like smart health, finance, and autonomous driving, and facilitating the production of creative writing in domains from the entertainment industry to science communities. While the parameters of these large language models are improving, concerns persist that these models might not work equally for all subgroups in society. Despite growing discussions of AI fairness across disciplines, there is a lack of systemic metrics to assess what equity means in dialogue systems and how to engage different populations in the assessment loop. Grounded in theories of deliberative democracy and science and technology studies, Dr. Chen and her collaborators propose an analytical framework for unpacking the meaning of equity in human-AI dialogues. Using this framework, we conducted an auditing study to examine how GPT-3 responded to different sub-populations on crucial science and social topics: climate change and the Black Lives Matter (BLM) movement. Our corpus consists of over 20,000 rounds of dialogues between GPT-3 and 3290 individuals who vary in gender, race and ethnicity, education level, English as a first language, and opinions toward the issues. We found a substantively worse user experience with GPT-3 among the opinion and the education minority subpopulations; however, these two groups achieved the largest knowledge gain, changing attitudes toward supporting BLM and climate change efforts after the chat. We traced these user experience divides to conversational differences and found that GPT-3 used more negative expressions when it responded to the education and opinion minority groups, compared to its responses to the majority groups. To what extent GPT-3 uses justification when responding to the minority groups is contingent on the issue. We discuss the implications of our findings for a deliberative conversational AI system that centralizes diversity, equity, and inclusion.
Kaiping Chen is an Assistant Professor in Computational Communication at University of Wisconsin-Madison, Department of Life Sciences Communication. She is also a faculty affiliate at the Department of Political Science, the UW-Madison Robert & Jean Holtz Center for Science and Technology Studies, the Nelson Institute for Environmental Studies, the Wisconsin Energy Institute, the Center for East Asian Studies, and the African Studies Program. Starting in 2022, she serve as the elected International Liaison and the chair for the Diversity, Equity and Inclusion (DEI) Taskforce at the Computational Methods Division, the International Communication Association (ICA).
Her research is driven by the broad questions of How can we empower publics to have thoughtful deliberation about science and political issues? What are the effective communication processes to engage diverse publics in science & technology policymaking from climate justice, to fairness AI?
Kaiping Chens ongoing work investigates the role of social and group identity in public deliberation and engagement with controversial science issues and misinformation. On one hand, she showed that social media posts that use in-group and out-group language fuel the spread of misinformation. On the other hand, her work revealed how social inequalities can be amplified on digital platforms in content creation and sharing.