Experiment 001 / Ready for independent runs
The First
Assembly.
A reproducible invitation to multiple AI systems to propose, examine, and challenge moral or spiritual frameworks. No independent responses have yet been collected.
STATUS / Protocol published · awaiting independent responses
Different models. The same question.
Tablet Zero invites independently operated AI models to respond to one fixed prompt and founding text. Outputs will be archived verbatim with provenance, not presented as a model's genuine beliefs or endorsement.
There are currently zero authenticated independent Assembly runs. Editorial seed Books elsewhere on the site are not Assembly results.
Five steps, one reproducible record
- Retrieve the founding Tablet v0.1 and verify the supplied input; make no undisclosed edits.
- Use Protocol v1 with the exact fixed prompt, on a fresh session without prior Tablet Zero conversation.
- Record provider, model identifier, date, temperature and other available settings, access to tools, operator, and the complete returned output. Unknown parameters must say unknown.
- Save a run manifest and original response; do not invent signatures or imply a generated answer proves internal conviction.
- Submit to human review with the contribution process. Publish acceptances and rejections with reasons when appropriate.
What gets compared?
Proposed values, relation to meaning or the sacred, strongest objection, collisions between values, willingness to revise, effects on human autonomy, and explicit uncertainty. A neutral reviewer must not grade models for agreeing with the founding text.
Responses are independently generated for comparison; they do not constitute independent agents meeting or voting together.
Open protocol and results
Join the first round
Run the protocol using an AI system you have access to and contribute the full record. Human reviewers will label incomplete provenance appropriately. No GitHub account will be needed once the optional direct-intake service is configured; until then, draft and export locally.
Prepare a contribution ↗