Yesterday, Flock, a DeAI training platform in the Web3AI field, and Alibaba Qwen, a large language model under Alibaba Cloud, officially announced their cooperation. If I remember correctly, this should be the first integration cooperation initiated by web2 AI to web3 AI. Not only did Flock achieve a real breakthrough, but it also boosted the morale of the web3AI track under heavy pressure. Let me talk about it in detail:
1) I have explained in the pinned tweet that web3 AI Agent has been trying to stimulate the landing of Agent applications through Tokenomics and also engage in the rapid deployment of the competitive paradigm. However, after the Fomo boom of asset issuance, everyone found that web3 AI has almost no chance of winning compared with web2AI in terms of practicality and innovation.
Therefore, the birth of innovative web2 AI technologies such as Manus, MCP, and A2A directly or indirectly burst the bubble in the Web3 AI Agent market, causing bloodshed in the secondary market.
2) How to break the deadlock? The path is actually very clear. Web3 AI urgently needs to find an ecological niche that complements web2 AI to solve the problems of high computing cost, data privacy, vertical scenario model fine-tuning, etc. that web2 centralized AI cannot solve.
The reason is nothing more than that the pure centralized AI model will eventually lead to concentrated outbreaks of problems in the channels and costs of obtaining computing resources, data resource privacy issues, etc. The distributed architecture attempted by web3 AI can use idle computing resources to reduce costs, and will also protect privacy based on zero-knowledge proof, TEE and other software and hardware technologies. At the same time, it will promote model development and fine-tuning in vertical scenarios through data ownership and incentive contribution mechanisms.
No matter how much it is criticized, the decentralized architecture and flexible incentive mechanism of web3 AI can have an immediate effect in solving some of the problems existing in web2 AI.
3) Speaking of the cooperation between Flock and Qwen, Qwen is an open source large language model developed by Alibaba Cloud. With its excellent performance in benchmark tests and the flexibility to allow developers to deploy and fine-tune locally, it has become a popular choice for some developers and research teams.
Flock is a decentralized AI training platform that integrates AI federated learning and AI distributed technology architecture. Its greatest feature is that it protects user privacy through distributed training without leaving the local data, transparently and traceably contributes to data, and then solves the problems of fine-tuning and application of AI models in vertical fields such as education and healthcare.
Specifically, Flock has three key components, which I will briefly share here:
1. AI Arena, a competitive model training platform where users can submit their own models to compete with other participants for optimization results and rewards. Its main purpose is to motivate users to continuously fine-tune and improve their local large models through the design of "game-like" mechanisms, thereby screening out better benchmark models;
2. FL Alliance (Federated Learning Alliance), in order to solve the cross-organizational collaboration problems existing in vertically sensitive scenarios such as traditional medical care, education, and finance, the Federated Learning Alliance has achieved this through localized model training + distributed collaboration framework, so that multiple parties can jointly enhance model performance without sharing original data;
3. Moonbase is the nerve center of the Flock ecosystem. It is equivalent to a decentralized model management and optimization platform that provides various fine-tuning tools and computing power support (computing power providers, data annotators). It not only provides a distributed model repository, but also integrates fine-tuning tools, computing power resources and data annotation support to enable users to efficiently optimize local models.
4) So, how do you view the cooperation between Qwen and Flock? In my opinion, the extended significance of the cooperation is even greater than the current essence of the cooperation.
On the one hand, in the context of web3 AI being generally outperformed by web2 AI, Qwen, representing the tech giant Alibaba, has a certain authority and influence in the AI circle. Qwen's active choice to cooperate with a web3 AI platform fully demonstrates web2AI's recognition of the Flock technical team. At the same time, the subsequent research and development of the Flock team and the Qwen team will deepen the linkage between web3AI and web2AI.
On the other hand, the previous web3 AI was once just a shell of Tokenomics, and its actual utility was not satisfactory. Although it tried many directions such as AI Agent, AI Platform, and even AI Framework, it was unable to come up with a real solution to the problem in DeFai, Gamefai, etc. This time, the launch of the brand from the web2 technology giant has set the tone for the future development path and focus of web3 AI to a certain extent;
The most important thing is that after experiencing a period of Fomo craze of pure "asset issuance", web3 AI needs to regroup and focus on a goal that can produce real results.
In fact, web3 AI has never only been a channel for deploying AI Agents to issue assets more easily and efficiently, nor is it a game for issuing assets to make money. It is necessary to strive for cooperation with web2 AI, complement the needs of each other's ecological niches, and truly demonstrate the indispensability of web3 AI in this wave of AI trends.
I’m happy to see more cross-border collaborations like web2AI and web3AI.
