/ projects · 03
Situated participation & AI consciousness
My UCL dissertation. Nagel famously asked what it is like to be a bat. I argue that's not a yes-or-no question, and that the graduated version changes how we should talk about large language models.
The argument
The debate around machine consciousness usually inherits Nagel's framing as a binary: either there is something it is like to be a system, or there is nothing. The dissertation pushes back on the binary itself. Phenomenological access, I argue, is graduated: systems participate in experience-like states to different degrees, depending on how they are situated in and coupled to a world.
Why LLMs
Applied to large language models, the framework reframes the question. "Is it conscious?" invites confident answers in both directions and evidence for neither. Asking instead where situated participation places a system on the gradient, and what kind of evidence could move it, turns a shouting match into a research programme.
Where it sits among the projects
This is the philosophy end of the PPE-with-data-science degree: the same interest that drives the machine learning projects (what can you actually claim from the inside of a model?) pointed at the hardest version of the question. No live demo on this one; some arguments don't fit in a widget.