Which Rules AI Systems should follow

Democratic Inputs to AI

OpenAI, Inc., is launching a program to award ten $100,000 grants to fund experiments in setting up a democratic process for deciding what rules AI systems should follow, within the bounds defined by the law.

I originally intended to participate but then backpedaled, as my idea is not that innovative and would not fit within the given time constraints (just 3 months from design to reporting).

If anyone is interested I can share the document (in private as this forum does not allow me to post links). Maybe something useful can be recycled from it (e.g. structure, metadata, examples).

Contrary to OpenAI’s example, my proposal would not have used AI much (save possibly for few supporting roles) because I believe AI is still too immature and liable to manipulations (e.g. training bias, dark patterns), centralized control and conflicts of interest.