We are at a critical inflection point.
The design choices being made now about how AI systems express uncertainty and interact with human judgment will shape whether these tools enhance human intelligence or quietly erode it.
The window for influence is narrow.
My work focuses on what it would take to get this right.
I argue that AI systems can be individually helpful while collectively deskilling entire professions: narrowing the questions communities ask and the values they debate.
The remedy lies in participatory infrastructure: tools designed not just for individual use but to support collective refinement over time.
At the Centre for Data Futures, which I founded at King's College London, we are building this participatory infrastructure across the entire AI lifecycle.
-At the point of data generation, the data empowerment clinic addresses what I call 'the missing profession of the 21st Century': students learn through advising groups or institutions interested in creating some data empowerment structure while a peer learning network facilities horizontal learning across different settings and geographies.
-At the point of deployment, we are notably piloting participatory interfaces that enable communities (whether professional groups like GPs, judges and educators, or civic organizations and neighbourhoods) to iteratively shape how AI systems communicate uncertainty in their domain.
I am the Inaugural Jeff Price Chair in Digital Law at King's College London.
I am also the founding Director of the Centre for Data Futures and a visiting professor at the Centre for Language AI research at Tohoku University (Japan).
My work has always been animated by a commitment to bridge the gap between theory and practice. This has led me to launch or be involved in a variety of ethics and public policy initiatives.
I am currently working on agency-enhancing uncertainty communication features for LLMs deployed in morally loaded contexts. I am also considering the social sustainability of the data ecosystem that makes generative AI possible.
Sylvie Delacroix