Knowledge about Artificial Intelligence (AI) and appraisal ratings: How people assess AI-generated images.

How to cite

Hauser, David; Crede, Ann-Kathrin; Gehrlein, Jonas; Bieberstein, Frauke von. 2019. "Knowledge about Artificial Intelligence (AI) and appraisal ratings: How people assess AI-generated images.." AEA RCT Registry. AEARCTR-0005065. https://doi.org/10.1257/rct.5065-1.2000000000000002

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

David Hauser* , Ann-Kathrin Crede , Jonas Gehrlein , Frauke von Bieberstein

* Primary investigator

Registry ID
AEARCTR-0005065
DOI
https://doi.org/10.1257/rct.5065-1.2000000000000002
First registered
Last updated
Status
In development
Trial dates
2019-10-10 – 2020-03-31
Countries
Private
Registry record
http://www.socialscienceregistry.org/trials/5065

Abstract

Through an online experiment on Amazon's Mechanical Turk, we examine how people assess images generated by artificial intelligence (AI) in order to gain more insights into how people perceive AI. By using three different information sets at the beginning of the survey, subjects are confronted with images generated by AI and rate them according to liking, emotion, creativity, and hypothetical willingness to pay. Finally, the evaluations of the AI-generated images of the treatment groups will be compared with a control group that receives no information.

Intervention

In a between-subject design, participants are asked to rate six images all created by artificial intelligence (AI) regarding liking, emotion, creativity, and hypothetical willingness to pay. We aim to examine whether subjects assess images generated by an AI differently when subjects:
i.) have no information how an image was created;
ii.) know that an image was created by an AI;
iii.) know that an image was created by an AI and also receive basic information about how the AI creates an image;
iv.) know that an image was created by an AI and also receive information about the person and team who programmed the AI.

Experimental design

Between-subject design.

Design details

Randomization method
randomization done in office by a computer.
Randomization unit
Individual.
Clusters
0.
Observations
In total around 560 participants.
Arms
Around 140 participants per treatment.
Primary outcomes
- Evaluation of the general liking of the images on a scale from 0 to 100. - Evaluations about experienced emotions, whether the subject experienced that the picture is communicating with him/her and the creativity and of the images on a 7-point Likert scale. - Hypothetical willingness to pay for the images. - Basic demographical metadata.

Cited by

No citation data available for this registration yet.

Keywords