Human judgment as a design variable in AI-enabled decision systems

Artificial intelligence is changing organisational decision making as capabilities increasingly move towards autonomous decisions and actions. The central challenge has moved beyond discussion of whether humans or machines make better decisions. It is now focused on designing systems that combine AI with human judgement to improve decision quality. This article argues that human judgement in AI-enabled systems should be designed, allocated and evaluated at the system level, rather than treated as a binary safeguard. Human involvement can be configured according to where judgement enters the process, when intervention occurs, how much authority humans retain, and what form of judgement is required. This perspective shifts the unit of analysis from the individual decision maker or AI model to the human–AI decision system. This article highlights emerging risks involved in the expansion of accountability for AI-enabled decisions. It also identifies the expertise paradox where AI automates the work through which humans actually develop the experience needed to supervise autonomous systems. This creates an important research agenda to help design intelligent decision systems and the role of the decision maker itself.