Work

Too Clear to Question: AI Summarization and Overestimation in Managerial Decision Making

Type: article
Abstract: Generative AI summaries are increasingly used to compress complex organizational information for managerial decision making. While often framed as productivity tools, summaries are not neutral reductions of evidence; they selectively emphasize, omit, and reorganize information before judgment occurs. This paper examines whether AI summarization weakens managerial decision quality while increasing decision makers’ overestimation of their own performance. Drawing on decision support research, dual-process theory, and overconfidence theory, we argue that fluent AI summaries may encourage heuristic processing and create an illusion of understanding. We further theorize that relevant industry experience attenuates this effect by improving domain schemas and metacognitive calibration. We propose a randomized experiment in which working adults complete an information-rich managerial decision task with or without an AI-generated summary. The study aims to contribute to information systems (IS) research by reframing AI-assisted decision making as a joint problem of performance and calibration.
Institution: McGill University
Language: English
Open Access status: green
OpenAlex ID: W7222992012
Cites: 40
Cited by:
Related to: 10