Why should AI framework standards move towards chaining?

Tác giả gốc: Hạo Thiên (X: @tme l0 211 )

Many people still don’t understand why I have been calling on AI framework standard project parties to move towards “chaining”? Perhaps it is because during the first two rounds of bull and bear markets, Chain infra carried too many expectations, and it is finally time to reach the era of AI Agent “application”, and everyone feels daunted by the “chain”. However, if AI Agents want to make more reliable autonomous decisions and communicate and collaborate, they must move towards “chaining”.

Currently, the popular ELIZA, ARC, Swarms and other frameworks are still in the concept stage. They cannot be falsified and return to zero, nor can they be confirmed and go up sharply. They are basically in the cradle stage where the valuation cannot be quantified. This is the first hurdle for issuing assets on Github. It is necessary to find the possibility of landing the outlined framework and vision in order to gain unanimous recognition from the market.

If we take a closer look at frameworks such as ELIZA, ARC, and Swarms, whether they are for extreme single AI agent performance optimization or multi-AI agent interactive collaboration frameworks, they essentially have to sort out a set of traceable logic and rules for AGI large model API calls.

After all, the data is off-chain, the reasoning process is difficult to verify, the execution process is opaque, and the execution results are uncertain.

From a short-term perspective, TEE provides a low-cost, highly feasible off-chain Trustless implementation solution that can accelerate the application of AGI to the autonomous decision-making of AI Agents. From a longer-term perspective, it also requires a set of on-chain consensus to assist in becoming more reliable.

For example, ELIZA wants to build an AI Agent autonomous private key hosting solution based on its framework, using @PhalaNetworks TEE secure remote authentication capability to ensure that AI-Pools execution code is not tampered with before calling the private key signature. However, this is only the first small step in the direction of TEEs effect on AI Agent.

If we can put the complex preset execution logic into the Agent Contract and let the Phala chains Validators participate in the verification, a chain based on the chain consensus to constrain TEE execution details will be opened. At that time, the positive flywheel of AI Agent driving TEE demand and TEE driving chain empowerment will start to run.

The logic makes sense. TEE can ensure that private keys are invisible, but how to call private keys, based on what preset rules to call, how to trigger risk control emergency response, etc. In the short term, it can be handed over to open source code base to achieve transparency, but in the longer term, doesn’t it have to rely on a decentralized verification consensus to verify and confirm in real time?

Vì thế, chaining can accelerate the AI Agent framework towards practical application and also bring new incremental opportunities to Crypto infra .

The direction is already very clear. For most people, it is important to find and bullish the earliest chained AI Agent framework and the earliest old chain that supports AI Agent. This is the Alpha under the new trend of AI Agent.

Liên kết gốc

This article is sourced from the internet: Why should AI framework standards move towards chaining?

Related: A quick look at 4EVERLAND, a comprehensive Web3 cloud service platform

By Alex Xu, Research Partner at Mint Ventures 1. Introduction Recently, 4 EVERLAND, a Web3 comprehensive cloud service provider that has been established and operated for more than 4 years, finally held a TGE. It started with decentralized front-end hosting services and gradually expanded to Web3 gateways, customized decentralized storage solutions, RaaS (Rollup as a service), blockchain RPC and AI RPC services, forming a product matrix centered on Web3 development needs. In this article, the author will investigate some key issues of the project to provide readers with an intuitive impression of the project, including: The projects core products and business content, and main customer groups Business data of the project Fundraising, Key Partners, and Token Model This article is the authors interim thinking as of the time of publication.…