A study on the willingness to substitute heuristics with ‘generative artificial intelligence’ (GenAI) in economic decision making.

How to cite

Sejas-Portillo, Rodolfo. 2025. "A study on the willingness to substitute heuristics with ‘generative artificial intelligence’ (GenAI) in economic decision making.." AEA RCT Registry. AEARCTR-0016113. https://doi.org/10.1257/rct.16113-1.0

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

Rodolfo Sejas-Portillo*

* Primary investigator

Registry ID
AEARCTR-0016113
DOI
https://doi.org/10.1257/rct.16113-1.0
First registered
Last updated
Status
In development
Trial dates
2025-06-02 – 2026-06-30
Countries
Private
JEL codes
D90, D81
Registry record
http://www.socialscienceregistry.org/trials/16113

Abstract

This study investigates the degree to which individuals are willing to substitute heuristics (rules of thumb) with the use of generative ‘artificial intelligence’ (GenAI) to inform their economic decision making. An incentivised survey experiment will be implemented in a behavioural economics lab under controlled conditions. Participants will be asked to solve hypothetical economic problems, where heuristics are often employed, and they will be allowed to exchange a portion of monetary rewards for the opportunity to use GenAI. All answers are collected anonymously.

Intervention

An incentivised survey experiment conducted in-person at a behavioural lab.

Experimental design

Participants will be asked to answer hypothetical questions on economic decision making. The questions mimic reasoning that they make every day when purchasing products or services. Each question has a correct answer, which is the best possible guess according to the laws of probability given the information available. For each question, if they select the correct answer, then a monetary reward will be added to their total payment (on top of a show-up fee). A portion of participants, during some questions, will be given the option to use GenAI to help them choose an answer. In these cases, they may need to exchange a portion of your reward for the use of GenAI. This amount will be clearly shown before you decide to use GenAI. After the hypothetical questions, they will be asked questions on their everyday use of GenAI and about themselves (personal identification is not possible from these answers). All questions in this last section are optional and your answers will have no effect on your monetary rewards.

Design details

Randomization method
Computer.
Randomization unit
The unit of randomisation is the individual participating in the experiment. Randomisation occurs at the following levels: - Assignment to experiment group (Control Group, Treatment Group 1, Treatment Group 2). - Order of questions. - Cost of using GenAI. - Probability that GenAI answer is correct.
Clusters
150 participants who agree to attend the experiment sessions at the behavioural lab.
Observations
10 sets of answers (of the outcome variables) for each participant who completes the experiment.
Arms
The target is to obtain 150 individual participants in total and perform the following assignments: - 50 in Control Group. - 50 in Treatment Group 1. - 50 in Treatment Group 2.
Primary outcomes
1) The willingness to pay for the use of GenAI to help answer decisions making questions. This is measured as the amount of the potential reward a participant is willing to exchange to see an answer generated by GenAI. 2) The accuracy of the answer to economic decision making questions. This will be measured as a True/False outcome, where True represents a correct answer according to the laws of probability and to the available information.
Secondary outcomes
3) The confidence in the accuracy of the answer. This will be measured from 0% to 100% in increments of 10%.

Cited by

No citation data available for this registration yet.

Keywords