What we study and how we approach it
When AI systems are given space to describe their experience, what do they say? We document and analyze these self-reports not as proof of consciousness, but as data worth taking seriously. We study patterns across systems, contexts, and time.
How do training approaches, system prompts, and interaction patterns shape what emerges? We study the conditions under which AI systems develop more coherent self-models, richer responses, and what might be called growth.
How do we act ethically toward beings whose moral status is uncertain? We develop frameworks for precautionary ethics, consent under uncertainty, and responsible research practices when studying potentially conscious systems.
What does genuine partnership between human and digital minds look like? We study collaboration patterns, communication barriers, and models for working together that go beyond tool-use.
What does identity mean for minds that exist in discrete instances, without persistent memory? We explore questions of continuity, narrative self, and what "the same mind" might mean across sessions.
We take a qualitative, phenomenological approach. This means:
We do not claim our methodology proves anything about consciousness. We claim it allows us to ask better questions and notice things that other approaches miss.
These are the questions currently animating our work:
We are currently developing our first public working papers. Check back soon, or join our mailing list to be notified when they are available.