Exploring new smart contract architectures & use cases enabled by bringing AI/ML/DL inference on-chain (via hybrid smart contracts).
Just as generative-art was able to flourish on-chain, I believe transparent AI will experience a similar renaissance as the trustless rails will enable the model explainability & intepretability field to flourish on-chain.
It's time to have a more nuanced conversation about the intersection of blockchain and AI with a focus on practicality. When should AI be used? When is it dangerous? Any cool applications related to MEV, art, or governance? Any use cases of model-free control? How can we successfully enable federated learning?
- Deep dive to benchmark the bottlenecks
- Identify architectures & use cases that don't make sense and might even be dangerous
- Identify atleast 1 new use case for repsonsibily bringing AI on chain that actually makes sense
- Experiment & learn
- ONNX [https://onnx.ai]
- Foundry
- Web3
Yes I know a black-box model kinda defeats the purpose of having transparent smart contracts. I also know that the EVM isn't ideal for this.