Gary Marcus argues that the AI industry should pause development of generative AI systems until safety concerns are adequately addressed. He contends frontier labs are building unreliable, uncontrollable systems they themselves acknowledge could pose catastrophic risks, while government oversight remains ineffective. Marcus advocates a public boycott of generative AI tools as economic leverage to force companies to prioritize safety, explicitly distinguishing his concerns from extinction-level fears in favor of nearer-term dangers like AI-generated pathogens and infrastructure hacks.
