Consumer Responses to Disclosing AI Involvement in Advertising Content Creation

How to cite

Meng, Juanjuan; He, Xinhao; Huang, Huaqing. 2026. "Consumer Responses to Disclosing AI Involvement in Advertising Content Creation." AEA RCT Registry. AEARCTR-0017641. https://doi.org/10.1257/rct.17641-1.0

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

Juanjuan Meng* , Xinhao He , Huaqing Huang

* Primary investigator

Registry ID
AEARCTR-0017641
DOI
https://doi.org/10.1257/rct.17641-1.0
First registered
Last updated
Status
In development
Trial dates
2026-01-15 – 2026-04-30
Countries
China ()
Registry record
http://www.socialscienceregistry.org/trials/17641

Abstract

This study will examine: (i) whether there will be differences in the quality of advertising copy written by independent humans, through human–AI collaboration, and by independent AI; and (ii) how consumers will respond to the disclosure of AI involvement in content creation. We plan to address these questions using a two-stage experimental design. In the first stage, we will collect advertising copy for suitcase (a durable good) and package tour service (service product). Based on uniform product information and specified writing topic, we will recruit participants to write advertising copy. Participants will be randomly assigned to either an independent human writing group or a human–AI collaboration group. In independent AI Group, copies were autonomously generated by ChatGPT. In the second stage, we will recruit participants through an online survey platform to evaluate the advertising copies, either with or without disclosure of authorship. We will measure participants’ willingness to interact with the content and their willingness to purchase the product. This design will allow us to examine differences in copy quality under non-disclosure, as well as the effect of AI disclosure, distinguishing between disclosure of independent AI creation and disclosure of human–AI collaboration. Finally, we plan to explore the mechanisms underlying the disclosure effect, focusing on curiosity, AI preferences and etc.

Experimental design

For Copy Writing:
Given uniform product info and specified writing topics,
Human Group:copies were independently written by employees of the company;
Human-ChatGPT Group: copies were collaboratively created by employees and ChatGPT;
ChatGPT Group: copies were autonomously generated by ChatGPT.

For Copy Rating:
the copy is presented with or without disclosure.
The wording of disclosure is:
This copy is written independently by human;
This copy is written collaboratively by humans and ChatGPT;
This copy is written independently by ChatGPT.

Design details

Randomization method
For people we recruit for copywriting, the randomization between "Independent Human Group" and "Human-AI Collaboration Group" is done in office by a computer. For copy rating in survey, the participant is randomized to different version of surveys by the survey platform.
Randomization unit
Individual
Clusters
No cluster. same as below.
Observations
Copy Collection: around 100 people * 4 copies per person= around 400 copies Copy rating: 600-800 people * 6 copy rating per person =3600-4800 observations
Arms
Copy collection: around 100 people, half in independent Human Writing Group; half in Human-AI Collaboration Group. Copy rating: 600-800 people, 300-400 for undisclosed survey; 300-400 for disclosed survey.
Primary outcomes
rating in the survey: including willingness to interact, willingness to purchase.
Secondary outcomes
curiosity measure and preference measure

Cited by

No citation data available for this registration yet.

Keywords