Author: Decentralised.Co
Compiled by: TechFlow
If artificial intelligence requires cloud services, then Web3 artificial intelligence requires Web3 cloud services.
@eigenlayer and AI have been the hottest topics in crypto over the past year. In this post, we’ll explore their intersection and some of the projects that are innovating in this space.
What is AVS?
First, we need to understand the Active Verification Service (AVS) on EigenLayer.
Think of EigenLayer as a marketplace for security and computing power.
Blockchain and other cryptographic protocols such as bridging rely on decentralized node operators to process transactions. These node operators are responsible for maintaining the current state of the network and processing incoming transactions. To verify a transaction, a majority of node operators must agree on its validity. Therefore, the greater the number of nodes, the more secure the network is.
New protocols often face a cold start problem when building a strong base of node operators. Operators are usually incentivized through the protocol’s native token. However, in the early stages, these tokens may have limited value due to the lack of a strong node network.
To solve this problem, the team may offer more tokens to incentivize node operators, but this may lead to high inflation and dilution of token value, which is not ideal. Moreover, in the early stages, the small number of nodes will also bring security and centralization risks.
EigenLayer solves this problem by helping any blockchain service, called an Active Validation Service or AVS, bootstrap Ethereum-backed security. The protocol consists of operators that specialize in providing computation and security. Users assign ETH or liquid staked ETH to these operators, who then validate one or more AVSs.
If the operator performs their duties, AVS will reward them, and they will distribute these rewards to depositors. If the operator fails to perform their duties, their stake will be slashed.
By having a common set of operators verify multiple services and governed by a standard economic layer, EigenLayer simplifies the launch of projects that rely on distributed nodes for security. This proposal has attracted a variety of projects including data availability solutions, bridges, oracles, and ZK processors.
AI
In the past two years, artificial intelligence has become the focus of the technology world, attracting the attention of entrepreneurs, investors, and users. This enthusiasm has naturally spread to the crypto field. According to @_kaitoai , artificial intelligence has become the most talked-about topic in all crypto fields in the past 12 months.
In the context of blockchain, operators are actually computers. When validating Rollups, they accept incoming transactions, process them, and output new states. However, if operators can provide hardware such as GPUs, SSDs, and ZK Provers, this input-processing-output model can be extended to any distributed computing operation. Therefore, EigenLayer can be regarded as a Web3 distributed cloud service provider.
Today, most AI processing is done in the cloud—from hyperscale cloud providers like AWS to specialized cloud providers like Lambda and Coreweave. These services support model training and reasoning, so EigenLayer, as a Web3 cloud, is a natural fit for Web3 AI applications.
Let’s look at some real-world examples.
Ritual
Currently, most users and developers access AI services through APIs from centralized cloud service providers. However, this status quo presents several problems, including loss of privacy, questionable computational integrity (how can you be sure that the response comes from the model you requested?), and potential for censorship.
In contrast, smart contracts run in a highly secure, transparent, and trusted environment. There are cases where smart contracts need to interact with AI services, but it is computationally infeasible to run any AI process on-chain. Existing cloud service providers are also unable to serve smart contracts because it would undermine their trust assumptions.
@ritualnet is solving this problem by building an open, privacy-first, censorship-resistant and verifiable AI layer designed for blockchain AI services. Their first product, Infernet, allows smart contracts to request AI model inferences with proofs of computational integrity. In the future, Ritual plans to expand by creating a sovereign chain, Ritual Chain, to provide more powerful features such as fine-tuning and training AI models.
Ritual Chain will be built as an AVS on EigenLayer. Operators with specialized hardware (such as GPUs) will execute AI queries for the chain. A decentralized set of validators will provide high availability and censorship resistance, as each query will be handled by multiple operators. In addition, these operators will also provide basic security for the Ritual Chain itself.
OpenLedger
A few weeks ago, we discussed the data challenges in AI and how blockchain protocols can play a role in solving them. While we recommend reading the entire article, the most important issue we highlighted is the centralization of AI data. Platforms that own valuable data strike high-value deals worth millions with well-funded companies, while limiting access to smaller startups and research institutions.
@OpenledgerHQ aims to provide a solution by creating an “AI sovereign data blockchain”. OpenLedger provides AI teams with:
High-quality annotated data to ensure effective training and accuracy
Reinforcement Learning and Human Feedback (RLHF) service for enhanced models
Tools for evaluating the accuracy, reliability, and security of AI models
OpenLedger is also building AVS on EigenLayer. While the exact implementation details have not yet been fully disclosed, we can make some reasonable guesses. In order to build a distributed, highly available data layer, the chain's nodes need a lot of fast memory. EigenLayer operators are well suited to provide this, as well as basic computation and security services.
Sentient
@sentient_agi announced an $85 million seed round earlier this month, attracting the attention of some of the top investors and operators in the crypto space. Their goal is to create an “open AGI development platform.” What exactly does that mean?
Currently, most of the top AI models are closed source and controlled by a few powerful organizations. This control is unhealthy for one of the most important technologies of our time. To this end, a growing open source movement is emerging, where the weights (configuration) of the model are open to anyone, allowing them to run the model on their own hardware or fine-tune it to their specific needs.
However, while open source models are vital, it is difficult for their creators to profit from them. Once the weights are public, anyone can host, modify, tweak, and create services based on them without sharing any revenue with the original model creators. This fundamental mismatch in incentives could slow the pace of open source AI development.
Sentient's goal is to bring "property rights" to AI development. It wants to create technology that enables researchers and developers to monetize AI models while keeping them open and secure. When developers use models created by Sentient, they can ensure the model's validity, just like with open source models. However, they need to compensate the creator of the model by paying for inference.
Sentient is built on Polygon CDK technology and serves as an AVS on EigenLayer. Although Sentient's specific use of EigenLayer has not been fully disclosed, we can speculate that its approach may be similar to Ritual. This may involve operators providing the computing resources required for inference as well as the security of the chain.
In a blog post last year, the EigenLayer team mentioned AI reasoning as one of 15 potential unicorn ideas that could be built as AVS. Apparently, many teams believe this potential is real. While it is still early days for EigenLayer and the Web3-AI space, the intersection between them is natural. If AI needs cloud services, then Web3 AI needs Web3 cloud services.
The projects we mentioned are just the first wave of preliminary experiments. We expect more projects to emerge.
