Polyhedra launches EXPchain as an application of artificial intelligence, explaining the necessity of AI model on-chain and the decentralized Zk proof generator

The interoperable infrastructure Polyhedra, after experiencing a decline in currency price and losing the $ZK code competition to zkSync, has recently launched a revitalized effort to build a "universal chain for AI," known as EXPchain. It has proposed the concept of Proof of Intelligence (PoI), creating an immutable and trustworthy Block chain for artificial intelligence models. Whether the successful transformation combining zk and AI is worth looking forward to.

Traditional AI supervision involves sensitive data, zkML becomes a new solution

The official definition of EXPchain is a Block chain protocol designed specifically for scalable, verifiable, and privacy-focused artificial intelligence applications. As the "Internet of Everything chain for AI", EXPchain integrates zero-knowledge machine learning (zkML) and a new Proof of Intelligence (PoI) framework. Key innovations include the efficient zk proof system Expander, as well as the integration of zkML into traditional AI workflows and the developer-friendly zkPyTorch toolkit.

Artificial intelligence plays an increasingly critical role in various industries, from using facial recognition to unlock smartphones to AI-driven loan applications and medical diagnoses. These technologies bring enormous potential as well as challenges. For example, how to ensure that AI systems operate fairly, accurately, and safely? How to protect sensitive data without compromising transparency and accountability?

In addition, governments around the world are also working on regulating AI, such as the EU's AI Act and the AI Risk Management Framework of the National Institute of Standards and Technology (NIST) in the United States. The problem with traditional methods is the need to disclose proprietary models or sensitive data, which leads to trade-offs between security, privacy, and trust.

Zero-knowledge machine learning (zkML) provides an alternative solution, the characteristics of zero-knowledge proof can protect data and model privacy while achieving mathematical verification of AI systems. Polyhedra introduces the interoperable protocol EXPchain based on zkML technology, which not only takes into account AI behavior and compliance norms but also is scalable and secure verification.

Technical debt continues to expand, AI trading process on the chain is beneficial for accountability

A study pointed out that the technical debt in the United States will reach $2.41 trillion in 2022. In addition, a study by PricewaterhouseCoopers (PwC), one of the world's four major professional consulting firms, also pointed out that by 2030, AI is expected to contribute as much as $15.7 trillion to the global economy.

As AI scales, it may exacerbate the expansion of technical debt. In this regard, the business column Raconteur has questioned whether companies are prepared to bear the cost of AI failure. AI failures include incorrect outputs, data leaks, and cyber attacks. In addition to economic losses, these errors also often cause harm to individuals.

For example, incorrect data output may lead to machine misjudgment or biased decision-making. Therefore, it is necessary to ensure that every element of AI-driven transactions, from data input to model output, is verifiable and accountable. Addressing these risks is crucial while unleashing the full potential of AI. This is where EXPchain, AI real-time verification of Block chain, comes into play.

Three major technological innovations: Can Polyhedra solve the problem of zk proof generator?

Technological innovation includes Expander, ExPos, and zkPyTorch

Polyhedra: Expander is currently the world's fastest zk-prover

Polyhedra provides data including:

Processing VGG-16 images on a single-threaded CPU takes only 2.2 seconds

Single-threaded CPU processing Llama-3.1 8B requires 150 seconds for each token

Performance is four orders of magnitude faster than previous data.

These developments have significantly reduced the cost and latency of AI verification, supporting various applications such as privacy inference to model review. Expander also aligns with Vitalik Buterin's zk ultimate vision.

Layer 2 is mainly divided into Optimistic Rollup and zk Rollup. For most zk Rollup public chains, the generation of ZKP proofs is a bottleneck, and companies must deploy powerful machines with TB memory to process the large number of transactions in ZKP. In a previous paper by Tiancheng Xie, the former technical director of Polyhedra, and Jiaheng Zhang, the chief scientist, they discussed a new solution that uses fully decentralized ZKP to improve the scalability of zk technology.

ExPoS: Expanded Proof of Stake

ExPoS is a proof of stake mechanism developed for the zkML technology in EXPchain, which verifies the behavior and compliance of AI applications without revealing proprietary model data. In simple terms, it uses the zkBridge technology from Polyhedra to unify and connect all the proof of stake mechanisms on the Block chain into a cohesive staking network.

zkPyTorch: Developer-Friendly Toolkit

zkPyTorch automatically converts PyTorch operations into zk circuits, bridging the gap between traditional AI development workflows and zero-knowledge machine learning (zkML). This integration allows developers to use familiar tools while significantly reducing the time and complexity of deploying zk-enabled AI applications.

zkML can complete LLM verification under the premise of privacy

EXPchain's core lies in zero-knowledge machine learning (zkML), which supports encrypted verification of AI models, achieving security and accuracy throughout the entire machine learning lifecycle, including:

Verifiable reasoning: proving the output of artificial intelligence without revealing the model or data.

Model review: Verify the fairness and compliance of performance based on the test set.

Training Verification: Ensure compliance with the protocol without disclosing sensitive inputs.

zkML Specific applications include:

Add digital watermarks to Large Language Models (LLMs). Digital watermarks are tiny and imperceptible features embedded in the text generated by LLMs, used to identify whether the text is generated by a specific model, and to prevent content forgery and misuse.

Ensure the compliance of the model, such as compliance verification in financial institutions.

Achieve secure multi-party computation in privacy-focused industries.

The zkML digital watermark of EXPchain can now be used to verify large language models such as Llama-3.1 8B.

Polyhedra's chief cryptographer has a strong background, promoting the PoI artificial intelligence proof chain

EXPchain can be seen as a Proof of Intelligence (PoI) that creates an immutable and trusted Block chain for artificial intelligence models, verifying their origin, authenticity, and ethical compliance. This framework protects intellectual property and ensures transparent accountability by cryptographically linking the source and performance of each AI model to verifiable records on the chain, providing unprecedented transparency for AI-driven ecosystems.

When it comes to the mastermind behind all of this, we have to mention Zhenfei Zhang, the Chief Cryptographer of Polyhedra. Previously, he has worked for industry leaders such as Algorand, Espresso, Ethereum Foundation, and Scroll, and he enjoys a considerable reputation in the field of cryptography. The article 'ZEN: an optimized compiler for verifiable zero-knowledge neural network inference' discusses verifiable machine learning.

This article Polyhedra launches EXPchain for artificial intelligence applications, analyzing the necessity of AI model chaining and the earliest appearance of distributed zk proof generators in Chain News ABMedia.