On June 7, at the "2025 Web3 x AI Innovation Forum" jointly organized by the Student Blockchain Association of Tsinghua University and Polyhedra, Hu Yuncong, assistant professor of Shanghai Jiao Tong University and co-founder of Polyhedra, delivered a keynote speech "Reshaping the Trust Boundary: A New Paradigm of Trusted Outsourcing Computing Based on ZK". Tiancheng Xie, Ph.D. from the University of California, Berkeley, and CTO of Polyhedra Co-Creation, delivered a keynote speech on "AI+Web3: The Prospect and Future of ZKML".
Among them, Xie Tiancheng mentioned: "The comparison between the complexity of verifying liquidity and computation is not on the same level. Specifically in the application of AI models, the primary value of ZKML (Zero-Knowledge Machine Learning) lies in verifying service quality. For example, in trading bots, users can verify the operational quality of the robot through ZKML.
Secondly, with the growth of AI Agents and their sensitive permissions being grantable, ZKML can effectively ensure the security of AI agents, meaning that users can act as verifiers to establish a trust system between humans and AI agents. Furthermore, ZKML can also provide data marketplace functionality, auditing training data capabilities, etc., which will play an important role in the subsequent on-chain AI construction.
The content is for reference only, not a solicitation or offer. No investment, tax, or legal advice provided. See Disclaimer for more risks disclosure.
Polyhedra co-founder Xie Tiancheng: ZKML will play a key role in verifying AI service quality and establishing a trust system.
On June 7, at the "2025 Web3 x AI Innovation Forum" jointly organized by the Student Blockchain Association of Tsinghua University and Polyhedra, Hu Yuncong, assistant professor of Shanghai Jiao Tong University and co-founder of Polyhedra, delivered a keynote speech "Reshaping the Trust Boundary: A New Paradigm of Trusted Outsourcing Computing Based on ZK". Tiancheng Xie, Ph.D. from the University of California, Berkeley, and CTO of Polyhedra Co-Creation, delivered a keynote speech on "AI+Web3: The Prospect and Future of ZKML".
Among them, Xie Tiancheng mentioned: "The comparison between the complexity of verifying liquidity and computation is not on the same level. Specifically in the application of AI models, the primary value of ZKML (Zero-Knowledge Machine Learning) lies in verifying service quality. For example, in trading bots, users can verify the operational quality of the robot through ZKML.
Secondly, with the growth of AI Agents and their sensitive permissions being grantable, ZKML can effectively ensure the security of AI agents, meaning that users can act as verifiers to establish a trust system between humans and AI agents. Furthermore, ZKML can also provide data marketplace functionality, auditing training data capabilities, etc., which will play an important role in the subsequent on-chain AI construction.