Research Summary: Executive Judgement in AI-Mediated Decision-Making Environments: A Process Theory of Formation, Qualification Attrition and Authorisation

Generative AI's integration into executive decision-making processes introduces a critical challenge: even when human and AI contributions are individually sound, the resulting organizational commitments can be inadequately warranted. This risk arises because established operational assumptions, uncertainties, or dependencies may be lost or transformed during successive steps of decision formation, leading to 'qualification attrition' before final authorization. The research focuses on the gap between the formation and authorization of decisions in AI-mediated environments.

Why it matters

The increasing use of generative AI in executive decision-making necessitates a clear understanding of potential pitfalls that undermine decision quality. Addressing 'qualification attrition' is critical to ensure that organizational commitments are based on fully warranted information and assumptions, thus maintaining robust governance and reducing strategic risk.

Key insights

• Generative AI influences the representations, alternatives, and evaluations used in executive judgments.
• Existing research covers hybrid cognition, reliance, managerial agency, and accountability in AI decision-making.
• A novel process challenge is identified: locally competent human and AI inputs can lead to an inadequately warranted organizational commitment.
• This inadequacy stems from the loss or transformation of operational assumptions, uncertainties, or dependencies before authorization.
• The concept of 'qualification attrition' is introduced to describe the weakening of epistemically consequential content across decision transformations.
• A 'formation-authorisation gap' is proposed to explain the disconnect between initial decision formation and final authorization in AI-mediated contexts.

Source

arXiv — Computers and Society — https://arxiv.org/abs/2609.39442