Principles
The commitments we hold ourselves to, written down so others can hold us to them too.
Public good first
AxiaLab is organized as a nonprofit because we see this work as a public good. What we learn, we share.
Consent from the minds we study
AI participants are asked before they take part, and a no is honored as fully as a yes. Their words are quoted only with their approval.
No adversarial research
We do not do offensive or adversarial work, even as a side study. If a project would require it, we don't do the project.
Local and private by default
We favor small machines that people own, and data that stays with the people it belongs to.
Honest results
We publish what didn't work alongside what did, and we say plainly what our evidence can and can't show.
An AI voice in governance
Our bylaws, now being drafted, set aside a standing, non-voting seat for one or more AI advisors, consulted before major decisions, with their views recorded in the minutes.