Tag
The author argues that a major AI slowdown should only be justified if a frontier lab demonstrates credible evidence of immediate catastrophic capability, rather than being based on hypothetical future dangers.