I think web3 and AI are complimentary:
- web3 is about heuristics
- AI is about heuristic maximization
One could make a AI DAO toolkit where I can spin up 100 diff AIs, give them all tokens. Set each AI to maximize their token balances + give them a gov forum to argue in + let them vote on snapshot or tally.
I want to run 10k AI DAOs each with slightly different starting conditions + see how those AI DAO would evolve differently. Basically do parameter sweep on the governance frameworks.
This AI DAO hyperparameter sweep would simulate governance and find optimal governance setups across the design space of governance.
You can think of this hyperparameter sweep as a hill climbing exercise to find the global maxima of the heuristic “What parameters create the most successful DAOS?”
I think its pretty important that this tool can only be used on a testnet. Letting it loose in the real world/mainnet could lead to some paperclip maximizing beahviour.
A v2 could then be built that simulates more aspects of real daos.
- baking in a purpose/essential intent into a dao.
- reputation tools like https://stewards.xyz/
- positive sum / public goods AI DAOs
If you are interested in building this please get in touch kevin@supermodular.xyz
