Ethereum co-founder Vitalik Buterin has suggested a temporary global "pause" on the availability of industrial-scale computing to mitigate potential risks associated with artificial superintelligence (ASI). In a Jan. 5 blog post, Buterin warned that ASI could arrive within as little as five years, posing unpredictable and potentially harmful consequences for humanity.

Buterin’s proposal builds on his philosophy of “decentralized and democratic defensive accelerationism.” This approach advocates for accelerating technologies that empower defense and decentralization while avoiding those that could concentrate power or cause harm.‍

The Risk of Superintelligent AI

Buterin highlighted the dangers of ASI, including the possibility of global disempowerment or catastrophic outcomes if the technology is misused. He acknowledged that while slowing down technological progress is often debated, it is difficult to enforce in a decentralized global environment. Instead, he proposed specific measures to mitigate risks.

One of Buterin’s ideas is a “soft pause” on industrial-scale computing power. This pause, lasting 1–2 years, would aim to slow the development of ASI and give humanity more time to prepare for its implications. He suggested implementing cryptographic mechanisms that require global authorization to continue operating high-powered computing devices.

This strategy, he argues, would target large-scale computing infrastructure while sparing consumer-level technology. It could serve as a temporary but effective measure to delay potential threats posed by ASI.‍

The Role of Decentralization

Buterin emphasized the importance of decentralized solutions in addressing the challenges posed by AI. He pointed out that decentralized technologies, such as zero-knowledge proofs and secure hardware, could enhance defenses against risks like cyberattacks and misuse of advanced AI.

As the race toward superintelligent AI accelerates, Buterin’s proposals add an important perspective to the ongoing debate about how to shape the future of AI responsibly.