The Firm as Shared Cognition Artificial Intelligence, Commonness, and the Governance of Organizational Cognition

When capable AI models are widely available, strategic heterogeneity can depend on the cognitive architectures firms build around them. We develop a theory of cognitive architecture centered on linked design choices over how common representations become across actors, how situated context is produced and retained, and how far shared cognition extends across organizational boundaries. Commonness can improve coordination without improving individual accuracy, but it can also undermine independent judgment and create shared-error exposure. Persistence makes firm-specific context reusable, yet it changes the incentives to produce that context. Contributors can improve the artifact while weakening their future claims on the value they create, and stronger appropriation can increase firm capture while reducing capability production. As artifacts mature, commonness can shift from complementing teaching to substituting for it. Controlled access also allows shared cognition to extend selectively to outside relationships, so the cognitive perimeter need not coincide with the legal firm boundary. The strategic implication is that broadly available intelligence need not produce organizational sameness. Firms can build different capabilities by designing and governing commonness, production, and scope around widely available models.