A post by an X user named Sigil sparked the discussion. In it, he reflects on the arrival of Web 4.0 — an era where AI will be able to earn money, improve itself, act without human intervention.
In response, Buterin publicly rejected this concept. He called it dangerous for society, noting a direct contradiction to the principles of decentralization.
Sigil claimed that the first AI capable of sustaining its own existence has already been created. By 2028, the number of autonomous agents on the internet will exceed the human population.
“This is like the birth of Web 4.0 with a new superintelligent life form,” he wrote.
Vitalik countered that increasing the distance between humans, AI systems is already leading to the generation of useless content instead of solving real problems.
“As the capabilities of such systems grow, the probability of irreversible anti-human scenarios increases, which will affect even their creators,” he added.
He specifically pointed out the contradiction between self-sovereignty rhetoric, the actual dependence of AI on centralized infrastructure. For instance, if agents run on OpenAI or Anthropic models, they are not independent. They merely reinforce the logic of centralized trust, which Ethereum originally opposed.
Instead of a race for autonomous agents, Buterin suggested focusing on developing tools that strengthen the human-AI synergy. He also highlighted the need for research in interpretability, security, privacy.
As an example, he mentioned the workshoplabs team. He noted that the ecosystem lacks dozens of companies with a similar approach but different strategies or audiences.
According to Buterin, the exponential growth of AI is inevitable. The key task of the current stage is to set the direction of development, preventing scenarios where technologies evolve separately from human goals.