Not sure if anyone is utilizing AI in the following use cases but wanted to throw out some ideas for potential development:
One of the significant issues within OpenGov is the large number of proposals, their length, and sometimes the complexity of the topics. @giotto has tried to address some of these using the concept of experts to discuss or otherwise summarize these proposals. While I believe in the short term and mid term this is a good idea, there should be the use of generative AI to ingest the content of these proposals and regurgitate them in a shorter simpler form that presents both the pros and cons of the proposal.
OpenGov is quite unique because it has solved the issues related to financial disclosures that many regulatory bodies want to see. The majority of crypto projects do not have an opensource governance software with on-chain verifiability that can execute payments and code autonomously with no human involvement. With that said, the next step on this journey would be to produce something similar to 10K and 10Qs, except for with the use of AI (which can be done with a click of a button). Generative AI could ingest all proposals passing, failing, passed, and failed within a given timeframe and create MD&A and disclosure equivalent forms in real time.
AI review of proposed code in governance (i.e., any user can run an AI against the proposed code to check for vulnerabilities).
Lastly, a project that can automate the creation of whitepapers for crypto projects. Or at least portions of the whitepaper again using generative AI.
AI has undoubtedly brought many benefits, particularly in simplifying tasks for everyone. However, when it comes to community proposals, it is essential to prioritize the voice of the community in deciding whether they support a project or not.
While AI can be incredibly helpful, we shouldn’t rely solely on it to make all decisions for us. After all, the strength of a community lies in its collective input and engagement. It is important to strike a balance and ensure that community members are actively involved in shaping the direction of projects.
Your idea is still very valid though, just my thoughts.
From my own attempts at using AI to explain Polkadot and Substrate concepts (or generate code), it is still lacking compared to using AI with Web2 tech stack concepts and code.
There would need to be an LLM specifically around Polkadot code, protocol, Substrate, etc. which then could be used for various use cases.
Also worth considering when you need AI to solve a problem in which the underlying protocol is at fault and should be improved upon to better work for humans – or better yet → change the underlying protocol to better work for AI
Never intended to use AI to replace the community. Quite the opposite actually. @giotto has brought this issue up on many occasions. The general user simply does not have the expertise to make informed decisions on many proposals (marketing, technical updates, etc.). And that assumes they have the time to read a whole proposal. Even so called experts may only be experts in one field (for example, someone might be good at marketing but not development). The world is highly specialized. Instead of relying solely on a circle of experts, we could also rely on AI to summarize proposals in shorter form with pros/cons, to read all comments in a proposal and summarize those (remove duplicates), and to provide month or quarter end summaries of what happened. These could act as checks against a smaller circle of people making decisions (again, augmenting rather than replacing).
Further, one of the true intents of Web3 is to automate in essence all back office functions of software based products while providing trust. Use of AI to further enhance disclosures would be the next logical step for OpenGov.
