Watch the critical role of verifiable machine learning (ML) in ensuring transparency and trust in AI, especially as AI adoption and complexities grow in data management, deployment, and monitoring. Using zero-knowledge (ZK) proofs, verifiable ML provides traceability directly at the model layer, vital for regulated industries like banking and DeFi.
Challenges with verifying AI outputs and existing system limitations were discussed, along with blockchain-based solutions to ensure secure off-chain computations for applications like risk modeling. Orion, an end-to-end verifiable ML platform built on Cairo, was introduced. Orion simplifies ML development by integrating familiar frameworks (e.g., PyTorch) while ensuring verifiability through ZK proofs without requiring deep technical knowledge.
