Vitalik: Afraid of AGI and ASI, supports building intelligence augmentation tools for humanity instead of creating superintelligent beings

ChainCatcher news, regarding recent AGI-related product developments in the field of OpenAI and AI, Ethereum founder Vitalik Buterin posted on the X platform, "My definition of AGI (Artificial General Intelligence) is: AGI is a sufficiently powerful artificial intelligence that, if one day all humans suddenly disappear and this AI is uploaded into a robotic body, it will be able to independently continue civilization.

Clearly, this is a very difficult definition to measure, but I think it captures the core of the intuitive distinction between 'the AI we are used to' and 'AGI' in many people's minds. It marks a shift from a tool that constantly relies on human input to a self-sufficient form of life. ASI (Artificial Superintelligence) is a completely different matter—my definition is when humans no longer add value to productivity in the loop (like in board games, we actually only reached this point in the past decade).

Yes, ASI scares me—even the AGI I defined scares me because it brings obvious risks of losing control. I advocate focusing our work on building intelligence-enhancing tools for humans, rather than creating superintelligent life forms."