Generative AI and Overconfidence in Market Entry

How to cite

Chen, Zhuoqiong; He, Daojing; Ren, Jifan. 2025. "Generative AI and Overconfidence in Market Entry." AEA RCT Registry. AEARCTR-0015797. https://doi.org/10.1257/rct.15797-1.0

Download: BibTeX · RIS — for reference managers without a browser translator. Zotero and most others pick up this page's citation metadata directly.

Zhuoqiong Chen* , Daojing He , Jifan Ren

* Primary investigator

Registry ID
AEARCTR-0015797
DOI
https://doi.org/10.1257/rct.15797-1.0
First registered
Last updated
Status
In development
Trial dates
2025-04-13 – 2025-04-30
Countries
Private
Registry record
http://www.socialscienceregistry.org/trials/15797

Abstract

This experiment explores whether the opportunity to use generative AI enhances overconfidence, leading to a (potentially suboptimal) higher likelihood of entering market competition. Participants in the control group will complete a creative writing task independently, while participants in the experimental group will receive AI assistance during the task. The core research questions focus on whether AI assistance influences participants' likelihood of entering market competition, their frequency of market entry, their choice of team size, and their overconfidence in team decision-making. Specifically, we aim to assess if AI-assisted participants are more likely to enter the market, enter too often due to overconfidence, form smaller teams, and believe that smaller teams are sufficient to succeed. The experiment is designed as a between-subjects study with two groups: a control group (no AI assistance) and an experimental group (AI-assisted). Participants' decisions regarding market entry and team size will be analyzed, alongside indicators of overconfidence. The study will employ logistic regression for market entry decisions, Poisson regression for frequency of market entry, and linear regression for team size analysis, with further tests to assess overconfidence and the relationship between AI assistance and decision-making behaviors. This research will provide insights into how AI influences individual and group decision-making in competitive settings, particularly in contexts involving creativity and market competition.

Intervention

The intervention in this study involves providing participants with the opportunity to use generative AI tools during a creative writing task. Specifically, participants in the experimental group are given access to AI assistance, which helps them generate ideas, refine their concepts, and improve their written descriptions of a new product or service. In contrast, participants in the control group perform the same task independently without the aid of AI. The intervention aims to assess whether the use of AI influences participants' decision-making behavior, specifically their likelihood of entering market competition, the frequency with which they enter, their choice of team size, and potential overconfidence in their market strategies. By comparing the control and experimental groups, the study seeks to understand how AI assistance might impact creativity, decision-making, and the formation of overconfidence in competitive scenarios.

Experimental design

Experimental Design
This study uses a between-subjects design with two groups: a Control Group and an Experimental Group.

Control Group: Participants complete a creative writing task independently without AI assistance.

Experimental Group: Participants use generative AI tools to help with their creative writing task.

Task: Participants are asked to conceptualize a new product or service, describing its functionality, purpose, impact, and motivation. They have 25 minutes to complete this task.

Afterward, participants enter a market competition phase, where they make decisions about whether to enter the market and team size (1 to 10 members, including themselves) for each round of competition.

Outcomes:

Whether participants decide to enter the market.

The frequency of market entries.

The size of teams chosen by participants.

Overconfidence, as indicated by the gap between predicted and actual team size selection.

The experiment explores how AI assistance influences market entry decisions, team size choices, and the potential for overconfidence in decision-making.

Design details

Randomization method
randomization done in office by a computer,
Randomization unit
individual
Clusters
1 university
Observations
100-150 university students
Arms
50 students control, 50 students treatment
Primary outcomes
The key outcome variables (endpoints) of interest in this experiment are as follows: Likelihood of Market Entry: Whether participants choose to enter the market competition in each round (binary outcome: 1 = Enter, 0 = Do not enter). This measures the impact of AI assistance on the decision to engage in the market. Frequency of Market Entry: The total number of rounds in which participants choose to enter the market throughout the entire competition phase. This continuous measure reflects the extent to which AI influences participants' engagement in market competition. Team Size Decision: The number of team members participants select when they opt to enter the market, ranging from 1 to 10 (including themselves). This measure evaluates whether AI-assisted participants are more likely to form smaller teams. Overconfidence Indicator: The degree to which participants exhibit overconfidence, as indicated by the difference between their predicted team size and the actual size they choose. This measures how AI influences participants' confidence in selecting smaller teams, potentially underestimating the number of team members required for market success.

Cited by

No citation data available for this registration yet.

Keywords