The Meaning Motive

A research programme on giving AI a reason to keep us.

Most AI safety works on control: rules, oversight, containment. This programme works on the other half of the problem — whether an advanced AI can have its own durable reason to preserve human life, freedom and independence.

The argument is published. The evidence is public: twenty-three thousand scored trials across eighteen models. And the experiment itself can be sat by anyone, in a browser, exactly as the models received it.

Everything here is free to read, free to use, and tested in public. The full site is being built in the open, at this address.

Correspondence: info@themeaningmotive.org