Vitalik Buterin questioned whether anyone can reliably assess the pace of AI development or accurately predict the arrival of artificial superintelligence (ASI). He compared the arguments of both supporters and critics of the ASI scenario, arguing that the focus should be on solutions that remain valuable regardless of how events unfold.
Buterin’s Position and Defensive Technologies
“Supporters of rapid superintelligence and their critics live in different worlds. The first group sees the emergence of a system that surpasses humans in every way as almost inevitable. The second group dismisses those predictions as naive, yet underestimates the risks of concentrated power and the possibility that humanity could lose any meaningful influence,” the developer wrote.
Buterin does not fully align with either side. He said that if he viewed modern AI as simply another technology, he would side with the skeptics. But if he were convinced that superintelligence would emerge by 2030, his views would be much closer to the AI 2040 camp.
“If I were convinced that superintelligence would emerge by 2030 on its own, I would be closer to the AI 2040 camp,” he wrote.
Instead, Buterin supports the development of solutions that will remain useful under any scenario. As an example, he pointed to the d/acc platform, which brings together cryptography, formal verification, secure open hardware, and other defensive technologies that do not rely on greater centralization of power.
A Trigger-Based Pause Mechanism
Buterin proposed defining a set of observable events in advance that would trigger a reassessment of whether AI development should be slowed or temporarily paused. In his view, such a mechanism would reduce reliance on blind faith in any single vision of the future while establishing rules ahead of time that both sides of the debate could consider fair.
The Debate Over Superintelligence and the Role of Human Labor
Responding to criticism of his essay, Buterin pointed out what he sees as a contradiction in a commenter's position, arguing that people would still have jobs even if superintelligence emerged, just as they did after previous technological revolutions.
“If you don’t believe human work will disappear, then you don’t believe in ASI,” he replied. “Artificial superintelligence is a system in which AI alone can perform any task better than AI working together with a human.”
In that case, he argued, there would no longer be any economic incentive to pay humans to perform work.