The purpose of our application is to conduct applied research into the usage of ChatGPT, LLama, or other LLM models in the context of cosmos community governance. Our thesis is that we can use the cosmos forum discussion data set to train a bot to make decisions on cosmos governance proposals. We can also give the bot a personality, such as Abraham Lincoln, Karl Marx, etc., by feeding the bot the public writings of these individuals, which embodies their values, preferences, and political economic thinking track records.
We believe we can make a fundamental contribution to large scale human governance systems, by removing the greed and self-interest from the congressman or senator and having only the intellectual knowledge as the basis of decision-making. In the cosmos context, this can solve the problem of bribing the validators, which we know has been a big problem to date.
- Build a testbed for LLM AI Governance
- Choose the optimum LLM model to act as a proposer for AI government
- Try to train the AI to acquire personality and memory characteristics
- Build a small social sandbox
- Attempts at various government systems based on AI proposers