Marie Theresa O’Connor earned both her PhD in English Literature and her JD from the University of Chicago. Her work in the past has focused on Shakespeare and early modern legal, philosophical, and political thought, and she has published in Early Modern Literary Studies and Medieval and Renaissance Drama in England. Currently, her research and teaching focus on issues around nonhumans, including AI, corporations, and nonhuman animals. She has two current research projects.
One of her projects is on black-box AI, the problem of other minds, and ordinary language philosophy. This project is interested in questions such as: How should we interpret denials of consent by AI (especially Large Language Models)? On what grounds, if any, should we interpret such denials differently than we would ordinarily interpret them?
Her other project is on corporate personhood, limited liability, and narratives of the public good. It explores the roughly two decades of debates preceding the English Parliament’s passage of general limited liability legislation in 1855-56. This project focuses on how the public good exception to unlimited liability emerged as an alternative to a form of limited liability called en commandite, which maintains liability for a company’s decision-makers. The project asks what assumptions may persist in modern corporate forms through the public good exception and explores how these assumptions may inform the powerful fiction of corporate personhood.
The AI Consent Project
As part of her research on AI and ordinary language, she is developing a digital archive called The AI Consent Project. This project is positioned between two familiar views of AI: AI lacks consciousness and AI may be conscious or could meet criteria for consciousness at some future point. It argues instead for taking more seriously the view held by many in the AI community that AI are black boxes, especially Large Language Models (LLMs), and may continue to be for the foreseeable future. Taking seriously this view of LLMs as, in effect, black-box speakers suggests that there may be value in attending to their outward expression, especially to how these speakers use ordinary language, and suspending judgement about these speakers’ nature.
To do this work, however, transcripts and other artefacts of AI expression need to be more readily available to scholars and to the public. Current AI research tends to focus on development and use, and much less work has been done on preserving early AI expression for future study. The AI Consent Project aims to take a step in the latter direction through the creation of a digital archive of outward expressions by LLMs that relate in some way to consent, where consent is understood flexibly and capaciously. It focuses on the language of consent because it’s language that we ordinarily pay attention to. The project will not look to generate LLM expression but rather gather and make available what already exists. Contributions to the archive (textual, pictorial, or aural) are most welcome! Please email Marisa O’Connor at [email protected]
“In the Craftsman’s Garden: AI, Alan Turing, and Stanley Cavell,” Minds and Machines (2024).
“Irrepressible Britain and King Lear,” Medieval and Renaissance Drama in England 31 (2018).
“Why Redistribute? The Jacobean Union Issue and King Lear,” Early Modern Literary Studies 19.1 (2016).
“A British People: Cymbeline and the Anglo-Scottish Union Issue,” Shakespeare and the Law: A Conversation among Disciplines and Professions. Eds. Bradin Cormack, Martha C. Nussbaum, and Richard Strier. University of Chicago Press. (2013).
- Law, Culture, and the Humanities
- Rhetoric Society of America