Vitalik: The risks of superintelligent AI should not rush forward

ChainCatcher message, Ethereum co-founder Vitalik responded on the X platform regarding discussions about AI, stating that superintelligent AI is very risky and should not be rushed forward, and that those trying to push it should be opposed. Do not build a $7T server scale.

A strong ecosystem of open models running on consumer hardware is an important hedge against the highly concentrated value captured by future artificial intelligence, where most human thoughts are read and mediated by a few central servers. The destruction risk of such models is also much lower compared to corporate arrogance and military forces.

Dividing artificial intelligence into "small" and "large" categories, exempting the "small" category and regulating the "large" category, seems reasonable in principle. Moreover, the scale of 405B is larger than consumer hardware (70B is not; I can run it). But the concern is that many current proposals will ultimately let everything slide into the "large" category.