Human AI Collaboration Preferences

How to cite

Ren, Yuqing; Ye, Teng; Zhang, Jundong. 2024. "Human AI Collaboration Preferences." AEA RCT Registry. AEARCTR-0014311. https://doi.org/10.1257/rct.14311-1.0

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

Yuqing Ren* , Teng Ye , Jundong Zhang

* Primary investigator

Registry ID
AEARCTR-0014311
DOI
https://doi.org/10.1257/rct.14311-1.0
First registered
Last updated
Status
In development
Trial dates
2024-09-16 – 2024-10-15
Countries
Private
Registry record
http://www.socialscienceregistry.org/trials/14311

Abstract

In this trial, we hope to study human preferences with how they collaborate with Generative AI tools in performing writing tasks. The study will be posted as a survey on online platforms, such as Amazon Mechanical Turk and Prolific. We will be varying the nature of the task and level of difficulty and observe how they affect the collaboration mode that humans choose to work with AI.

Intervention

We have two main sets of interventions: task type and task difficulty.

Experimental design

We will vary different types of writing tasks and the level of difficulty of the task and ask participants their preferences of whether and how to collaborate with a Generative AI agent, similar to Chat GPT. We will also collect data on the background and demographics of the participants to see if these factors also affect their preferences.

Design details

Randomization method
Randomization to experimental conditions will be done in Qualtrics.
Randomization unit
individual
Clusters
300-500 Amazon Mechanical Turk workers
Observations
300-500 observations
Arms
300-500 Amazon Mechanical Turk Workers as one cluster
Primary outcomes
Our main outcome is human preferences for AI collaboration modes, that is, whether to work with AI and how to work with AI.

Cited by

No citation data available for this registration yet.

Keywords