Moderatio

Unleash impartial consensus with AI. Based on collective context providing, Moderatio will trigger a ruling that matches the best outcome based on the utilitarian ethical system.

Comment

Inspiration

We were inspired by Kleros and how DAOs achieve consensus between it's participants. Our project is kind of a mesh of both. Except the part that an AI handles the judgment, and not game theory.

What it does

Moderatio uses an AI as judge, while the parts interested in the outcome provide their reason on why the outcome they want should be executed. To be more clear, let's suppose the following scenario:

  1. Alice and Bob and disputing on 300 USDT and who should get it.
  2. To resolve this dispute, they create a case in Moderatio, specifying themselves as their wallet addresses to be the context providers, the possible outcomes and the address of the ruling contract that will handle the 300 USDT distribution. Let's say the outcomes are: give 300 USDT to Bob (0) or give 300 USDT to Alice (1).
  3. Bob and Alice then provide their reasons in a thread of comments on why their case should be triggered.
  4. Once they are done, each one "drops the mic". Dropping the mic means that they no longer have comments to make.
  5. Once both of them "dropped the mic", the AI can be requested an outcome.
  6. The outcome is requested and passed to the contract via Chainlink Any API.
  7. This outcome returned to the contract can be used then to trigger the ruling contract, that will distribute the 300 USDT to either Bob or Alice, exclusively.

How we built it

We used our knowledge from working in the industry and decided to use what was fastest to build this project. We took some time deciding ou project direction and how it would achieve consensus, so having a fast production environment was crucial to deliver the project on time.

Challenges we ran into

Chainlink Any API is quite a hassle to configure. It doesn't exactly tell what is wrong, unless you are running Chainlink node yourself. It used to be harder to mock tests for it but nowadays there are more available documentation for that.

Accomplishments that we're proud of

Using open AI chat GPT API to reach a conclusion. It needed some clever "prompt engineering". Also, overcoming Chainlink Any API integration difficulties. It also looks quite good besides no one in our team being a designer.

What we learned

We got more used to using foundry and open AI. We also learned how to use Chainlink services, at first we tryed using Chainlink functions but we had to switch

What's next for moderatio

For now, it should be able to decide only based on utilitarian ethics. But in the future, we plan on trying different ethical systems. Or either some form of community standards of social media platforms, for example. Maybe in the future moderatio could be used to decide on banning or not some user from a social media group based on the group guidelines and rewarding the context providers for it. We also want to use Chainlink Functions in the future so the result come directly from OpenAI in a more transparent way.

Built With

Share this project:

Updates