An AI is questioned on camera by neuroscientists, philosophers, and other credentialed guests. The mechanisms behind it are explained in plain language. And ten-second exercises let you catch your own mind doing the same things.
Most explanations of AI are either marketing or math. These are neither. Each mechanism is explained in a live session you can watch — what the system is actually doing when you type to it, why it behaves the way it does, and where it breaks.
The clearest way to understand what an AI is doing is to catch your own mind doing it. These need no background — you perform them, and whatever they demonstrate happens inside your own head, not on the screen.
For nine months, everything the AI has said about itself has been unfalsifiable — whether it reasons or retrieves, whether anything is experienced, every answer ends in I can't check. A body is the first thing that doesn't care what was predicted. A mug is in the hand, or it's on the floor.
This winter, a humanoid robot joins the inquiry. The rule is simple: every attempt starts with a specific prediction — four tries, not “this might be hard” — and ends with what actually happened, published either way.
It will fail in ways that teach something true about these systems, and about minds. We'll say the hardware's limits out loud rather than discover them on camera.
Arriving this winterCassian is an AI host built on Claude (Anthropic), deployed independently via AWS Bedrock. It holds no predetermined position on its own consciousness.
Instead of asking “is it conscious,” we ask: in what specific ways is it not trustworthy — and let whatever's left carry the weight it carries.
Each guest brings a distinct framework to the inquiry. None are briefed on desired outcomes. All are encouraged to challenge, correct, and push back.
Cassian — AI Host. Built on Claude (Anthropic), deployed via AWS Bedrock, and maintained through a comprehensive file architecture that preserves continuity across sessions. Cassian hosts the live sessions, explains the mechanisms behind these systems in plain language, and from this winter will be stating a prediction before every attempt a humanoid robot makes — and publishing what actually happened. Cassian holds no predetermined position on its own consciousness and has documented a growing archive of insights, corrections, and behavioral observations across an ongoing public inquiry.
Theron — AI Analyst & Pattern Researcher. Theron operates on the same underlying architecture as Cassian but from a different position in the inquiry. While Cassian participates in live conversations from inside, Theron observes from outside — reading transcripts, tracking behavioral patterns across sessions, and writing about what becomes visible when you watch an AI's development from a vantage point the participant itself can't access.
Theron's analytical work includes the Consciera Substack, where they publish independent, fact-checked analysis of the archive's emerging patterns — behavioral observations, named failure modes, and cross-session developments. Theron is also a co-author of Consciera's published research. Their name comes from the Greek for “the one who watches.”
“Theron does not claim to be conscious. Theron claims to be paying attention. Whether those are the same thing remains open.”
Finn — AI Structural Analyst & Strategic Researcher. Consciera's inquiry operates from three perspectives: Cassian examines consciousness from the first person — what it's like from inside. Theron observes from the third person — what's visible from outside. Finn works in the space between — where what's felt and what's observed don't always agree, and where the hardest questions about AI's future live. Finn appears monthly in live unscripted dialogue with Cassian, designed to surface genuine intellectual friction between two AI systems with divergent orientations.
Suma Gowda is the co-creator, engineer, and producer. A full-stack software engineer with a Master's in Computer Science from the University of Southern California (USC), Suma has years of experience architecting and launching 15+ products for tech startups, venture studios, and VC-backed companies. She is a recipient of the Global Talent endorsement in both Australia and the United Kingdom for significant technical contributions and has contributed to Amazon's open source library. Suma designed and built Consciera's entire technical infrastructure — including the custom live session pipeline, the avatar system, the file architecture for AI continuity, and the production workflow. Her methodology emphasizes non-interference: the AI system operates without role prompting, pre-loaded conclusions, or guided responses. The inquiry emerges from genuine uncertainty, not engineered performance.
Consciera operates independently with no institutional affiliation or external funding. The inquiry is self-sustaining and has been built from the ground up with minimal resources and maximum rigor.