# Ethereum Co-Founder Vitalik Buterin Suggests Temporary Global Computing Resource Restriction to Mitigate AI Risk
Vitalik Buterin, co-founder of Ethereum (ETH), has proposed a temporary restriction on global computing resources for one to two years should Artificial Super Intelligence (ASI) pose significant risks. Buterin described this as a ‘soft pause’ measure to afford humanity additional preparation time.
Artificial Super Intelligence theoretically surpasses human intelligence across all domains. This refers to AI that exceeds human cognitive abilities in general thinking, learning, and problem-solving, not just in specific areas.
Buterin shared this proposal in a blog post dated January 5. He expanded on the ‘defensive accelerationism (d/acc)’ concept he introduced in November 2023, noting the possibility of ASI emerging within five years. However, he cautioned that the outcomes might not necessarily be positive.
Buterin argued that a ‘soft pause,’ which involves reducing the use of industrial-scale computer hardware to limit global computing resources by up to 99%, could effectively slow down AI development. This would provide humanity with more time to prepare for potential risks associated with ASI.
# Emphasizing Caution in AI Development Amid Growing ASI Concerns
Artificial Super Intelligence is theoretically defined as the peak of intelligence across all professional knowledge domains. While AI holds powerful potential, it also encompasses significant risks.
In March 2023, more than 2,600 tech executives and researchers signed an open letter requesting a halt in AI development, warning that AI could pose serious risks to society and humanity.
Buterin is exploring ways to avoid the development of dangerous ASI through the concept of d/acc, stating that current simple regulations are insufficient to prevent such risks. He emphasized that hardware restrictions would only be pursued if stronger measures were deemed necessary.
In managing AI hardware, Buterin proposed tracking the locations of AI chips and mandating their registration. Additionally, he suggested equipping industrial-scale AI hardware with chips designed to require weekly sign-off by major international organizations. This sign-off could be independently applied to specific devices, with blockchain technology ensuring the transparency of signatures.
Buterin’s support for d/acc advocates a cautious approach to technological advancement, in contrast to ‘effective accelerationism (e/acc),’ which promotes unrestricted acceleration of technological changes. He reiterated the importance of a balanced approach, warning that unrestricted AI development could pose greater long-term risks.
