Soma is the standard of care that artificial intelligence owes the people it decides for. Published in the open, free to read and use — a position, not a product.
Who gets hired, lent to, flagged, or seen — choices that always carried a duty of care now run through machines that feel none. Soma exists so that duty isn't lost in the handover. The founding note sets it out in full.
A machine can take over the decision. It cannot take over the duty of care that came with it.
Long before machines, anyone whose work could alter a life was held to a standard of care — attention, competence, honesty owed to the one with the least power in the exchange.
When the deciding moves to a model that screens, scores, ranks, or flags, that duty does not stay behind. It travels with the decision — and too often lands on no one. Soma names where it lands, and what it asks.
The decision must meet a real standard — sound, current, fit for the person it concerns — not merely whatever the model returned. A system that decides worse than a careful person is a hazard that was chosen.
The person has a right to know a machine decided, and to understand the grounds in terms they can act on. "Proprietary" is not an answer one is owed. Care that hides itself is not care.
Every automated decision needs a door back to a human with the authority to look again, and to overturn. A decision no one can revisit has not been governed — only imposed.
One does not deploy and walk away. The duty continues for as long as the system decides — watching for the person it begins to harm, and reaching them before the harm compounds.
"We will not be liable" and "they are alright because of us" are different sentences. Soma is built on the second.
Soma is a standard for our new shared reality.
Read it, cite it, disagree with it — carry it into your own corner of the world. And if you'd like to bring it into your teams, to write your own standard of care together, that's what the workshops are for.