CONCEPT

Judgment Specificity

The Williamsonian insight that AI is simultaneously despecifying execution capability—making it generic and commodified—while respecifying judgment, rendering evaluative and contextual competence more transaction-specific, organizationally embedded, and economically valuable than ever before.

Oliver Williamson’s central variable was asset specificity—the degree to which an asset loses value when redeployed outside a particular relationship. The higher the specificity, the more a governance structure is needed to protect the asset from the hazards of market exchange; the lower the specificity, the more markets can do the work. AI has introduced a bifurcation of unprecedented sharpness into the knowledge economy: execution capability is being despecified, and judgment capability is being respecified. Execution—the ability to write code in a particular language, to design an interface, to analyze data, to draft a document—was once highly transaction-specific: it took years to acquire, was difficult to substitute, and created the bilateral dependencies that justified long-term employment relationships. AI has made it generic. Any competent professional equipped with AI tools can now perform, at a level sufficient for most organizational purposes, tasks that once required specialists. The specificity has dropped, and with it the bargaining power of workers whose value resided primarily in execution. But judgment—the capacity to evaluate whether the output serves genuine need, to assess architectural soundness against organizational history, to make the call that no dataset fully informs—has not been despecified. It has deepened. It is more organizationally embedded, more contextually dependent, more irreplaceable than ever, because it is now the only scarce resource that the abundance of execution has not touched. The firms that understand this are reorganizing around judgment specificity; the firms that do not are discovering that the speed that eliminated the cost of building the wrong thing did not eliminate the cost of choosing the wrong thing.

Origin

The concept emerges directly from Williamson’s framework applied to the AI transition. Transaction cost economics predicts that when asset specificity declines for one category of activity, the governance structure shifts from hierarchy toward market. Workers whose value resided in despecifying execution face a genuine erosion of bargaining power: the market alternatives have expanded dramatically, the bilateral dependency has weakened, and the governance structure shifts accordingly. This is not a moral claim but a structural prediction, and it is already visible in the labor market reorganization following the AI transition.

The respecification of judgment is less immediately visible but more consequential. The capacity to evaluate whether AI-generated code will scale under load, to assess whether AI-assisted analysis has captured the relevant causal relationships or merely the superficial correlations, to determine whether AI-drafted strategy serves the organization’s actual competitive position—each of these is deeply embedded in organizational context. It depends on knowledge of particular customers, particular competitive dynamics, particular institutional histories. It cannot be generalized, cannot be outsourced to a tool, and as execution becomes abundant, it becomes the scarce resource around which organizational value concentrates.

The historical parallel Williamson’s framework makes precise: the Luddite framework knitters of Nottingham whose skill was genuinely valuable, genuinely hard to acquire, genuinely the product of years of practice. The power loom did not produce fabric of equal quality—but it produced fabric of sufficient quality at a fraction of the cost, and sufficiency, not superiority, is the threshold that determines whether a market shifts governance structures. The senior software architect who could feel a codebase like a pulse has experienced exactly this: not that his skill became less beautiful, but that it became less necessary at the execution layer, even as the judgment that accumulated alongside it became more necessary than ever.

Key Ideas

The bifurcation: execution despecified, judgment respecified. AI makes execution generic and judgment specific. This bifurcation is not gradual but sharp, and it affects different workers differently depending on where their value resided. Workers whose primary asset was execution face eroding bargaining power; workers whose primary asset is judgment face increasing organizational indispensability. The net effect is not the dissolution of the firm but its reorganization around a different category of specific asset.

Judgment cannot be accelerated. The geological metaphor from [YOU] on AI—every hour of debugging depositing a thin layer of understanding that accumulates over years into solid ground—is, in Williamson’s terms, a description of the investment process that produces transaction-specific human capital. That process cannot be shortcut without loss, because the specificity of the knowledge depends on the slow accumulation of contextual experience that no computational tool can replicate. AI skips the deposition and produces the surface—but it is the deposited layers, invisible beneath the surface, that constitute judgment.

The governance implication: concentrate human attention on judgment. Organizations that understand judgment specificity will build governance structures that concentrate human attention on the transactions where evaluative depth matters and delegate to AI the transactions where execution speed matters more. This is precisely what Williamson’s discriminating alignment hypothesis predicts: governance structures will align with transaction characteristics. The firm that continues to organize around execution costs that no longer justify hierarchical governance, while neglecting the judgment costs that now do, is answering the make-or-buy question with yesterday’s calculus.

Debates & Critiques

The primary debate is whether judgment really is distinct from execution or whether, as AI capabilities expand, judgment will also be despecified. If sufficiently capable AI can evaluate its own outputs, assess its own reasoning, and identify its own errors with the contextual sophistication that currently requires a human with organizational immersion, then the bifurcation Williamson’s framework currently predicts may be temporary. The current evidence cuts both ways: AI models do improve in their ability to detect their own hallucinations, but their errors become subtler rather than fewer as capability increases, requiring ever more sophisticated evaluation. The more capable the model, the more organizational context the evaluator needs to catch what it gets wrong. A second debate concerns the distribution of judgment specificity within organizations: whether it concentrates in a thin layer of senior decision-makers (exacerbating inequality) or whether the development of judgment can be democratized through apprenticeship and institutional structure, allowing a broader workforce to claim the organizational specificity that execution capacity no longer provides. Williamson’s framework is descriptive rather than normative here, but the stakes of the distributional question are substantial.

Origin

Key Ideas

Debates & Critiques

Related Entries

Further Reading