Math Olympian in Shadow of John Nash Tries to Solve Blockchain, AI Trust Dilemma

In mathematics, Jasper Zhang figures to be a sort of Zeus. He says he won gold medals at math olympiads in China and Russia, and it took him just two years to get a Ph.D. from the University of California, Berkeley.

Now he’s trying his hand at solving a key problem at the intersection of two of the fastest-growing but most complicated areas – blockchain and AI.

Jasper Zhang is a speaker at CoinDesk’s Consensus Festival, May 29-31, in Austin, Texas.

Hyperbolic, the two-year-old startup that Zhang leads focused on decentralized AI computing, said Thursday that it is introducing a protocol called “Proof of Sampling (PoSP),” aimed at addressing challenges with trust in decentralized AI networks.

Hyperbolic was co-founded in 2022 by Zhang and Yuchen Jin, who holds a Ph.D. in computer science from the University of Washington.

The concept for the new protocol was created in conjunction with researchers from Berkeley and Columbia University, according to the team. It combines math, computer science and economics, deploying “advanced sampling methods and game theory to incentivize integrity and minimize computational demands across decentralized networks,” Hyperbolic shared in a press release with CoinDesk.

Zhang, 28, said in an interview with CoinDesk that he sees PoSP as the next iteration of verification for decentralized networks.

“People in the beginning thought there’s only one way to do verification, which is with consensus,” Zhang said. “Later on people discover optimistic proving and then ZK proofs.”

Now there’s PoSP, he said, and it can not only be applied to AI, but also to rollups, a type of layer-2 blockchain, as well as so-called actively validated services (AVSs), which are protocols secured by restaking protocols like EigenLayer.

A research paper on the Proof of Sampling Protocol by Zhang and several co-authors was submitted on May 1 to arXiv, an open-access repository hosted by Cornell University for scientific papers that have not yet been peer-reviewed.

According to the paper, the design relies on a “pure strategy Nash Equilibrium.” That refers to a game theory concept attributed to the Princeton University-educated mathematician John Nash, who was the subject of the 2001 Oscar-winning film A Beautiful Mind, directed by Ron Howard and starring Russell Crowe.

Here’s a figure from the paper illustrating the architecture:

The Proof of Sampling architecture (Zhang et al)

As part of the release, Hyperbolic is introducing “spML,” an implementation of PoSP built specifically for AI verification.

“SpML leverages the foundational principles of PoSP to create a verification mechanism that is not only faster and more secure but also economically feasible,” Zhang said in the press release.

Now they just have to prove it works in practice.

Edited by Bradley Keoun.


Leave a Reply

Your email address will not be published. Required fields are marked *


Get latest news delivered daily!

We will send you breaking news right to your inbox

4Coinz ©. All rights reserved.