Lost in Conversation? Uncertainty, ‘value-alignment’ and Large Language Models

Professor Sylvie Delacroix
Wednesday, 27th November 2024 @ 12:30pm (GMT)

Abstract: Since LLMs can be used as conversational partners, the types of uncertainty they need to be able to convey are not limited to quantifiable, semantic or factual types. How LLMs relay morally-loaded unknowns will have a qualitative impact on the nature of future conversations. This talk not only draws attention to this hereto under-appreciated systemic risk. It also explores potential avenues towards a symbiotic learning environment where both the LLM and the user community continuously evolve in their understanding and communication of unquantifiable uncertainties.

About the presenter: Professor Sylvie Delacroix is the Inaugural Jeff Price Chair in Digital Law. She is also a fellow of the Alan Turing Institute. Her research focuses on the role played by habit within ethical agency, the social sustainability of the data ecosystem that makes generative AI possible and bottom-up data empowerment. The latter work led to the first data trusts pilots worldwide being launched in 2022 in the context of the Data Trusts initiative www.datatrusts.uk. The public policy dimensions of her work have led her to being invited to contribute to multiple policy initiatives. She has also acted as an expert for public bodies, such as the UK’s DCMS Department, and served on the Public Policy Commission on the use of algorithms in the justice system (Law Society of England and Wales). Her current work on agency-enhancing, participatory infrastructure and the communication of uncertaintyin the context of LLMs deployed in morally-loaded contexts has been funded by Omidyar Network. Previously, Professor Delacroix's work has been funded by the Wellcome Trust, the NHS, Mozilla Foundation and the Leverhulme Trust, from whom she received the Leverhulme Prize. Her latest book Habitual Ethics? was published by Bloomsbury in 2022 (open-access).