Paper Number
ICIS2025-2549
Paper Type
Complete
Abstract
This study investigates whether conversational generative AI (GenAI) systems, specifically ChatGPT, can reduce the impact of cognitive biases on human decision-making, e.g., by providing information and offering alternative perspectives, and whether a specific system prompt configuration can enhance this debiasing effect by inducing reflection. We conducted an online experiment with 434 participants, who made decisions before and after a single interaction with ChatGPT configured with either a standard system prompt or a reflection-inducing system prompt. The results indicate that interacting with ChatGPT, configured with a standard system prompt, significantly reduces the impact of cognitive biases on decision-making across all scenarios. However, the reflection-inducing prompt did not promote the debiasing effect on decision-making as expected. The findings of this study provide the first empirical evidence on the debiasing potential of ChatGPT, supporting the conceptualization of conversational GenAI systems as cognitive assistants that augment the limited cognitive capacities of human agents.
Recommended Citation
Wyszynski, Marc; Westermann, Jan; Weber, Sebastian; and Niehaves, Björn, "GenAI Improves Decision-Making: Revealing the Debiasing Capabilities of ChatGPT" (2025). ICIS 2025 Proceedings. 31.
https://aisel.aisnet.org/icis2025/gen_ai/gen_ai/31
GenAI Improves Decision-Making: Revealing the Debiasing Capabilities of ChatGPT
This study investigates whether conversational generative AI (GenAI) systems, specifically ChatGPT, can reduce the impact of cognitive biases on human decision-making, e.g., by providing information and offering alternative perspectives, and whether a specific system prompt configuration can enhance this debiasing effect by inducing reflection. We conducted an online experiment with 434 participants, who made decisions before and after a single interaction with ChatGPT configured with either a standard system prompt or a reflection-inducing system prompt. The results indicate that interacting with ChatGPT, configured with a standard system prompt, significantly reduces the impact of cognitive biases on decision-making across all scenarios. However, the reflection-inducing prompt did not promote the debiasing effect on decision-making as expected. The findings of this study provide the first empirical evidence on the debiasing potential of ChatGPT, supporting the conceptualization of conversational GenAI systems as cognitive assistants that augment the limited cognitive capacities of human agents.
When commenting on articles, please be friendly, welcoming, respectful and abide by the AIS eLibrary Discussion Thread Code of Conduct posted here.
Comments
12-GenAI