Artificial Intelligence and Human Creativity

How to cite

Djawadi, Behnud Mir; Fahr, René. 2025. "Artificial Intelligence and Human Creativity." AEA RCT Registry. AEARCTR-0016306. https://doi.org/10.1257/rct.16306-1.2

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

Behnud Mir Djawadi* , René Fahr

* Primary investigator

Registry ID
AEARCTR-0016306
DOI
https://doi.org/10.1257/rct.16306-1.2
First registered
Last updated
Status
In development
Trial dates
2025-07-04 – 2026-04-30
Countries
Private
JEL codes
C91, D91, O31
Registry record
http://www.socialscienceregistry.org/trials/16306

Abstract

We investigate the role of generative artificial intelligence (AI) on human creativity. Participants receive matchsticks (without heads) and are required to assemble a form within a 5-minute time frame. In addition to a control treatment with no assistance, the AI treatment allows partic-ipants to select among the following options: 1) no assistance, 2) AI assistance with preselected prompts or free prompts. At the end of the ex-periment, participants complete a questionnaire with socio-demographic questions and self-assessed items about creativity and their experience with AI in daily and professional life. The assembled forms are photographed and presented to a judging panel that ranks the art works using predefined criteria of originality and elaboration. The top 3 participants of each session receive monetary prizes of decreasing size. Among other objectives, we are interested in determining to what extent AI supports or impedes creativity, whether subjects who rate them-selves as uncreative assemble more creative forms with AI assistance compared to uncreative-rated subjects without assistance, and whether this difference is also observed among subjects who rate themselves as creative.

Experimental design

Participants receive matchsticks (without heads) and are required to assemble a form with-in a 5-minute time frame. In addition to a control treatment with no assistance, the AI treatment allows participants to select among the following options: 1) no assistance, 2) AI assistance with preselected prompts or free prompting. At the end of the experiment. participants complete a questionnaire with socio-demographic questions and self-assessed items about creativity and their experience with AI in daily and professional life. The assembled designs are photographed and presented to a judging panel that ranks the art works using predefined criteria of originality and elaboration. The overall score is the simple sum of the two dimensions of originality and elaboration, weighted equally. The top 3 participants of each session receive monetary prizes of decreasing size.

Design details

Randomization method
Randomization done in office by a computer by randomly scheduling when the baseline and when the treatment takes place.
Randomization unit
Individual
Clusters
We plan to collect for the baseline and the treatment between 50-70 participants each. As we are also interested how children perform in this task we also collect 50-70 children observation for baseline and treatment each. Instead of money they receive toys.
Observations
Between 50-70 adults and 50-70 children for baseline and treatment
Arms
50-70 per treatment arm (baseline + treatment)
Primary outcomes
The overall score (consisting of the simple sum of the single scores of the dimensions originality and elaboration) assigned by the judgment panel to participants' completed designs. Each dimension is weighted equally. The two scores in the single dimensions of originality and elaboratio are also used.

Cited by

No citation data available for this registration yet.

Keywords