Couldn't agree more. This episode perfectly frames the core veriability challenge in blockchain and AI. The contrast between SNARKs for deterministic proof and human-AI alignment in DocETL is crucial. It reveals trust's complex nature across these systems.
The parallels between Justin's trust minimzation in blockchains and Shreya's work on AI verifiability are fascianting. The idea that SNARKs could lower barriers to entry for verifying transactions while AI systems need similar statistical primitves for humans to check outputs really highlights how both fields are wrestling with the same core problem. It's intresting that both are using decomposition strategies to balance resource costs with acuracy.
Couldn't agree more. This episode perfectly frames the core veriability challenge in blockchain and AI. The contrast between SNARKs for deterministic proof and human-AI alignment in DocETL is crucial. It reveals trust's complex nature across these systems.
The parallels between Justin's trust minimzation in blockchains and Shreya's work on AI verifiability are fascianting. The idea that SNARKs could lower barriers to entry for verifying transactions while AI systems need similar statistical primitves for humans to check outputs really highlights how both fields are wrestling with the same core problem. It's intresting that both are using decomposition strategies to balance resource costs with acuracy.