Progress In the Decentralized AI Stack
summary
Panel argues that current generative agents are useful but limited—agents act like interns rather than autonomous managers; practical builds include Telegram bots with ~200,000 users and experimental crypto-native agent projects like Olas ecosystem. Security use cases pair AI with onchain safeguards, e.g. monitoring mempool and flicking a kill switch for harmful transactions. Verification relies on TEE, ZK proofs and research into FHE (notably Zama (FHE)), and debate centers on GPU markets vs. mining for funding and scalable inference.
AI-written summary · about
chapters (10)
- 0:00Introduction
- 1:17AI and crypto sentiment
- 3:03Decentralized AI use cases
- 7:46On-chain vs off-chain agents
- 10:16AI and surveillance concerns
- 15:37Agents as interns or managers
- 21:22Centralization risks
- 27:19Verified AI models
- 28:42GPU allocation
- 31:00Q&A
related talks
27mFrom Theory to Real-World Impact
35mThe Fight for the Future of Artificial Intelligence
32mPanel Discussion: Trends in Generative AI
31mPanel: Trustworthy AI: Enhancing Human Agency and Machine Accountability in the Age of Agentic AI
15mBuilding AI Agents the Web3 Way
14mReliable Agentic AI, the Open Way
01h 42mBuilding Responsible AI Agents in Decentralized Open-Source Environments
32mHow do we build infrastructure for the agentic economy?