Value of AI Explanations

How to cite

Chan, Alex. 2025. "Value of AI Explanations." AEA RCT Registry. AEARCTR-0016777. https://doi.org/10.1257/rct.16777-1.1

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

Alex Chan*

* Primary investigator

Registry ID
AEARCTR-0016777
DOI
https://doi.org/10.1257/rct.16777-1.1
First registered
Last updated
Status
In development
Trial dates
2025-09-14 – 2025-12-31
Countries
Private
Registry record
http://www.socialscienceregistry.org/trials/16777

Abstract

The study measures why individuals are willing to pay (WTP) to see an explanation of an AI model’s prediction, distinguishing: (a) curiosity/consumption value, (b) sense of control when evaluating a prediction, and (c) instrumental value when participants hold private information that can be combined with an explanation to improve accuracy. This experiment documents a new context in which failures of contingent reasoning affect the valuation of information.

Intervention

Subjects have to make a prediction about whether a specific loan applicant repaid a loan. AI prediction was offered. We elicit subjects' willingness to pay for explanations for AI predictions using BDM. We vary timing of when the explanation can be seen (before or after decision), with or without private signal, and with or without tutorial on contingent reasonsing.

(This is referred to as secondary experiment in AEARCTR-0015581, where we stated that we will separately pre-register this experiment. Similar bot-detecion techniques stated in AEARCTR-0015581 will apply here)

Experimental design

Between-subjects 2×2 design, crossed with a within-subjects manipulation

Design details

Randomization method
Qualtrics in-build randomizer
Randomization unit
individual
Clusters
1000 individuals
Observations
1000 individuals
Arms
1000 individuals, 250 per arm
Primary outcomes
Willingness to pay for explanation (in $$)
Secondary outcomes
prediction about the loan status (Defaulted or Repaid)

Cited by

No citation data available for this registration yet.

Keywords