Predictive systems increasingly participate in organizational judgments about performance, risk, allocation, diagnosis, and future action. Their organizational significance, however, extends beyond predictive capability. When predictive output is treated as unusually authoritative, employees may face social, procedural, or epistemic costs for challenging it, potentially weakening the flow of discrepant local knowledge that organizations require for correction and learning. This Original Propositional Model Article develops a bounded theoretical account of that possibility. It distinguishes technological capability from legitimate authority, voice from silence, psychological safety from consequential influence, and algorithmic confidence from the proposed social signal of algorithmic certainty. Building on these distinctions, the article introduces the proposed constructs of predictive authority, algorithmic certainty signaling, defensive conformity, and anomalous-knowledge suppression, and integrates them in a proposed Dissent-Suppression Model. The model theorizes that predictive authority may reduce consequential dissent when authoritative framing, interpersonal risk, responsibility diffusion, opacity, or dependence make contradiction costly, thereby limiting the entry of anomalous knowledge into organizational review and weakening specified learning processes. The article further develops falsifiable propositions and conditional countermeasures centered on contestability, independent judgment, protected anomaly escalation, and learning-preservation review. These mechanisms are not presented as empirically validated, universal, or deployment-ready. Their operation should vary with task characteristics, expertise, psychological safety, system quality, work design, hierarchy, and level of analysis. The contribution is a testable organizational theory of when predictive systems may become epistemic authority structures rather than neutral decision aids.
