A fraud-proof dispute game you can rig: tap any step of the computation trace to plant a fault, then watch the chain bisect proposer vs watcher commitments — six queries isolate the one divergent step, it gets re-executed on-chain, and the liar's bond is slashed.
Zero-knowledge proofs aren't the only path to trustworthy on-chain AI. Optimistic schemes trade latency for a 1000x cost reduction — here's how dispute games over inference actually work.