Governance shouldn't just be a set of rules we ask AI to follow.
It needs to be part of the engineering.
At Atlas Workshop, we developed our philosophy, Constitution and stewardship principles before writing the software.
The challenge since then has been turning those principles into working systems that separate AI reasoning from permissions, decision-making, memory, learning and human authority.
We've now implemented and tested a number of these governance mechanisms, with more engineering and independent validation still ahead.
The diagram below gives a simplified look at how governance flows through Atlas, without getting into the detailed software architecture.
I have also just launched a new page on our website, The Workshop Window, giving people a look behind the scenes at how were building Atlas.
I will update it periodically as the engineering, research and testing progress.
https://www.atlasworkshop-ai.nz/the-workshop-window
The goal is simple: build AI around the human, not the human around AI.
Thinking with, Never for.
#AtlasWorkshop #AIGovernance #HumanAI #AIEngineering #ResponsibleAI
Add comment
Comments