Paper Type

ERF

Abstract

Generative AI systems increasingly support expert decision making, yet their highly fluent and polished explanations may unintentionally weaken professional vigilance. This study examines how linguistic eloquence in generative AI outputs influences experts’ ability to detect reasoning errors in high stakes contexts. Using a between-subjects experiment, we manipulate the fluency of an AI-based decision support system while keeping its analytical content constant, including a predefined logical error. Drawing on Dual Process Theory and Cognitive Fit Theory, this study investigates whether stylistic fluency triggers heuristic processing, reduces cognitive effort, and encourages premature cognitive offloading. Expected contributions include extending existing cognitive theories by introducing linguistic cognitive fit, showing how stylistic alignment between AI language and expert expectations can create an illusion of reasoning quality that suppresses analytical verification, and informing the design of AI interfaces that maintain cognitive engagement in professional decision making.

Paper Number

1646

Comments

SIG HCI

Download

Share

COinS
 
Aug 15th, 12:00 AM

Eclipsing Expertise: Linguistic Eloquence as a Catalyst for Cognitive Passivity in Generative AI Interactions

Generative AI systems increasingly support expert decision making, yet their highly fluent and polished explanations may unintentionally weaken professional vigilance. This study examines how linguistic eloquence in generative AI outputs influences experts’ ability to detect reasoning errors in high stakes contexts. Using a between-subjects experiment, we manipulate the fluency of an AI-based decision support system while keeping its analytical content constant, including a predefined logical error. Drawing on Dual Process Theory and Cognitive Fit Theory, this study investigates whether stylistic fluency triggers heuristic processing, reduces cognitive effort, and encourages premature cognitive offloading. Expected contributions include extending existing cognitive theories by introducing linguistic cognitive fit, showing how stylistic alignment between AI language and expert expectations can create an illusion of reasoning quality that suppresses analytical verification, and informing the design of AI interfaces that maintain cognitive engagement in professional decision making.

When commenting on articles, please be friendly, welcoming, respectful and abide by the AIS eLibrary Discussion Thread Code of Conduct posted here.