Abstract
For artificial intelligence (AI) technology to impact society positively, the major AI companies must coordinate their efforts and agree on safe practices. The social legitimacy of AI development depends on building a consensus among AI companies to prevent its potentially damaging downsides. Consortia like the Partnership on AI (PAI) aim to have AI competitors collaborate to flag risks in AI development and create solutions to manage those risks. PAI can apply valuable lessons learned from other industries about how to facilitate collective action but do so proactively rather than after the fact. The Dynamic Capabilities Framework of “sensing, seizing, and transforming” provides a process map for the AI industry to create processes to reduce the risk of a major disaster or crisis.