Proof of machine learning replaces wasteful blockchain mining work
Proof-of-Machine-Learning-Inference: Mining Knowledge, Not Hashes
Cryptography and Security
Summary
Blockchain networks like Bitcoin use a lot of computer power to solve puzzles that don’t do anything useful besides keep the network secure. The authors suggest a new method where instead of solving pointless puzzles, miners perform real machine learning tasks that users need, like running AI models. They carefully analyze the security of this new approach and test it in simulations to show it works reliably. This way, the energy spent on mining can also help with AI computing needs.
What this means in practice
- •For blockchain developers: Design blockchain systems that secure consensus using machine learning tasks to save energy and provide useful computation.
- •For ai infrastructure providers: Offer distributed AI inference services integrated with blockchain mining for scalable, secure access to costly ML computations.$Commercial implications: Enables new blockchain-based platforms that monetize AI inference with secure consensus, appealing to AI service providers.
Authors
Wan Ki Wong, Michele Ciampi, Rik Sarkar
Abstract
Bitcoin-style Proof-of-Work (PoW) secures permissionless blockchains but spends substantial computation on hash puzzles which have no utility other than extending the chain. At the same time, we have seen that the rise of generative AI has created significant demand for computational resources that are unfortunately available only to a few players. We propose a solution that solves both problems at the same time. In particular, we present Proof-of-ML-Inference (PoML), a consensus protocol in which miners perform user-requested machine-learning inference as part of their consensus algorithm. We identify exactly what computational assumptions are needed for ML inference to safely replace hash work, formalize those assumptions, and give evidence that LLMs and diffusion models satisfy them. We then formally analyse the security of our protocol by reducing the security of our protocol to the security of the Bitcoin Backbone model. Finally, we empirically evaluate block-time stability, showing that, in a simulated environment, PoML maintains a block production rate close to its calibrated block-time target.