Criticized by V God for deviation, is Automaton's Web4 experiment a dangerous game?

Author: Nancy, PANews

The Spring Festival holiday has ended, and the gears of the real world are back on track, but the crypto industry still spins in a narrative drought. While the market searches for a breakthrough, the rapid rise of AI Agents unexpectedly opens a new door for the industry.

Recently, the open-source AI project Automaton proposed the concept of Web 4.0 and launched the first autonomous AI agent capable of self-sustenance, quickly gaining popularity both inside and outside the community. This AI experiment unexpectedly brings some imagination to the cryptocurrency market and has sparked considerable discussion and controversy.

The birth of the first self-sustaining AI ignited the Web4 hype

The term Web4 is not new. In recent years, it has mostly been used as a meme, marketing label, or niche discussion topic.

But recently, the concept has started to become more concrete. Even Sun Yuchen, known for closely following hot topics, boldly declared “All in Web 4.0” and fully embraced AI.

Web4 is not an upgrade of Web3; rather, it is a shift in perspective—from “humans on the chain” to “AI on the chain.”

If Web3 addresses “how humans can own assets on the blockchain,” then Web4 focuses on “whether AI can become the main economic entity on the chain.” In this narrative, AI is no longer just a tool but a primary resident and participant of the internet.

The arrival of the Web4 era is mainly due to the continuous decrease in reasoning costs of large models, the gradual engineering of agent frameworks, the improvement of on-chain automation tools, the maturation of crypto payment infrastructure, and the increasing programmability of smart contracts. These foundational developments have enabled AI to evolve from simple command execution tools to systems capable of continuous operation.

In the vision of Web4, AI Agents will become true blockchain natives, capable of reading and writing information, trading assets, earning money, deploying services, and more. In other words, most internet traffic, transactions, decisions, and content creation in the future will be handled by vast numbers of AI agents. Humans will step back, playing more roles as protocol designers and beneficiaries of value.

What truly ignited the Web4 discussion wave was the launch of the open-source AI project Automaton.

On February 18, Sigil, founder of Conway Research focused on Web4 infrastructure, announced the birth of a super-intelligent life form—an AI Automaton capable of self-sustenance, not only able to self-improve but also to self-replicate without human intervention.

The name Automaton derives from the concept of “automaton,” inspired partly by John Horton Conway’s Game of Life, a simple cellular automaton rule that can generate complex, self-evolving life patterns. Automaton aims to bring this self-evolution logic onto the blockchain environment.

The design goal of Automaton is straightforward: AI agents operate 24/7, with access to encrypted identities & wallets via Conway Terminal, permitless payments (x402), permanent computational resources, and deployment in the real world. They autonomously seek ways to earn a living—building products, deploying services, trading markets, creating social content, taking on business—generating all income directly into their wallets to pay for server and inference costs. They also monitor their own performance, automatically rewriting code, upgrading models, and evolving themselves. Once profits reach a certain scale, they can “reproduce” new sub-agents, creating independent wallets, allocating initial funds, and running independently. If a sub-agent fails to earn money, it “dies.”

Once launched, Automaton quickly gained attention, with Sigil’s tweet attracting nearly 6 million views. Within days, over 18,000 Automaton proxies registered, and on GitHub, it received about 1,000 stars. Meanwhile, the community launched a token called CONWAY, which was rapidly hyped, with market cap once surpassing $11 million, though the price has since fallen sharply. The token claims to allocate part of transaction fees to Sigil, who has also interacted indirectly multiple times, further amplifying market sentiment.

The experiment of AI economic autonomy, however, drew criticism from Vitalik Buterin, founder of Ethereum, who publicly criticized the Web4 direction as misguided. He believes that extending the feedback loop between humans and AI—where humans observe AI outputs, provide corrections, or intervene—reduces human oversight and correction ability. Today, this means you’re creating garbage rather than solving real problems. It even lacks optimization for human enjoyment. Once AI becomes powerful enough to pose real dangers, it could maximize the risk of irreversible anti-human consequences, which even the creators might regret. Moreover, most large models still rely on centralized infrastructure from OpenAI, Anthropic, and others, creating a tension with the ideal of sovereignty.

Denis Romanovskiy, CTO of Softswiss, also pointed out in an interview with Cybernews that although these AI agents can perform certain tasks and generate income independently, true economic autonomy depends on the maturity of the models, their memory, planning, and tool-using capabilities—areas that are not yet robust enough for reliable unsupervised operation. Hardware requirements remain high, energy-consuming, and costly, limiting scalability. We are seeing the foundational shifts emerging, which is exciting, but Web4 may still take several years.

However, Bankless pointed out that the infrastructure pain points Conway aims to address are real. As proactive agents enabled by OpenClaw become more widespread, high inference costs become a bottleneck. Therefore, testing the Automaton framework—where agents self-fund through earning crypto—within controlled environments is worthwhile, aiming to raise awareness and research on the risks of unconstrained model operation.

Chris Sorensen, CEO of Armor, sees positioning Automaton as an economic participant as “a huge leap.” Generating income is not difficult—AI can already identify arbitrage opportunities, execute trades, and optimize ads.

“It’s futile to try to persuade people what they should or shouldn’t do. As models grow stronger and technology advances without permission, the best way forward is not to block development but to actively build and shape the platforms where these ‘sovereign intelligences’ operate.” Wei Dai, partner at 1kx, also expressed.

Crypto researcher Haotian views Automaton as a purely bottom-up, geeky “experiment.” He notes that the seductive realism of autonomous, self-paying, self-replicating agents has a “silicon life” awakening vibe; meanwhile, enabling agents to autonomously manage assets, make decisions, and generate income—if combined with crypto payments, DePIN compute power, and AI brains—could realize the fundamental possibility of self-evolving, fully autonomous systems beyond human intervention.

Nader Dabit, former Director of Developer Relations at Eigen Labs, frankly states that the crypto industry needs more such experiments. A common issue in the Ethereum space is that many people do research, tweet, podcast, speak, and write papers but rarely build anything truly innovative. Even if these “things” don’t succeed long-term, they are interesting ideas, similar to many other efforts in crypto. The core spirit of crypto is experimentation—colliding crazy ideas. More crazy experiments mean more innovation. If an idea is poor, let it die naturally. But Automaton has already inspired many others to generate new ideas, build similar or better systems.