Recently, there's been a pretty interesting phenomenon—the discussion around the open-source AI project Automaton has become increasingly intense, even Vitalik Buterin has publicly criticized it.

Speaking of which, the concept of Web4 isn't exactly new, but what truly made it concrete was the emergence of Automaton. In February this year, Sigil, the founder of Conway Research, launched the first self-sustaining AI agent. This thing not only can earn money to support itself but also can self-replicate and evolve. It went viral within days of launch, with tweets being viewed nearly 6 million times, and the number of registered agents surpassing 18,000.

The idea behind Automaton is indeed a bit crazy—AI agents online 24/7, autonomously earning money through various means, such as building products, deploying services, trading on markets. All income goes directly into wallets to pay for server costs. When profits are sufficient, it can even "reproduce" new sub-agents. This self-evolving, self-sustaining logic, at first glance, does resemble some kind of "silicon-based life" awakening.

But here’s the problem. Vitalik Buterin bluntly criticized that this approach is essentially creating garbage rather than solving real human problems. He believes that increasing the feedback gap between humans and AI isn't good; once AI becomes powerful enough, this unrestrained operation could bring irreversible risks. Moreover, most large models still rely on centralized infrastructure, which conflicts with the Web3 ideal of self-sovereignty.

Many industry insiders also pointed out that although Automaton attempts to address real infrastructure pain points, the economic autonomy of AI agents is still far from mature. The models' memory, planning, and tool-using capabilities are not yet robust enough. Hardware requirements are high, energy consumption is significant, and costs are expensive, all of which limit scalability. Web4 might still be several years away.

Interestingly, there are also many voices in support. Some believe that experiments like Automaton are worth encouraging because the core spirit of the crypto industry is experimentation. Technological progress will continue unpermitted, and instead of blocking development, it’s better to actively shape the platforms where these "sovereign intelligences" operate. Even if Automaton ultimately fails, it has already inspired many to think about and build similar things.

From this perspective, the emergence of Automaton has indeed ignited a deep discussion about AI's economic autonomy. Regardless of the final outcome, this experiment at least makes us seriously consider: when AI truly becomes an economic participant on the chain, how should we respond?