What Skills Matter in the Age of AI? Evidence from a Responsible AI Curriculum in Rural China

Last registered on September 21, 2026
View Trial History

Pre-Trial

Trial Information

General Information

Title
What Skills Matter in the Age of AI? Evidence from a Responsible AI Curriculum in Rural China
RCT ID
AEARCTR-0019607
Initial registration date
September 09, 2026

Initial registration date is when the trial was registered.

It corresponds to when the registration was submitted to the Registry to be reviewed for publication.

First published
September 21, 2026, 8:07 AM EDT

First published corresponds to when the trial was first made public on the Registry after being reviewed.

Locations

There is information in this trial unavailable to the public. Use the button below to request access.

Request Information

Primary Investigator

Name
Yiming Xia
Affiliation
RIEM,Southwestern University of Finance and Economics
Contact Primary Investigator

Other Primary Investigator(s)

PI Name
Xin Meng
PI Affiliation
Australian National University
Contact Investigator
PI Name
Sen Xue
PI Affiliation
Renmin university of china
Contact Investigator
PI Name
Guochang Zhao
PI Affiliation
Southwestern university of finance and economics
Contact Investigator
PI Name
Yiming Zhang
PI Affiliation
Australian National University
Contact Investigator

Additional Trial Information

Status
In development
Start date
2026-09-15
End date
2026-12-31
Keywords
Education
Additional Keywords
generative AI; human capital; AI literacy
JEL code(s)
C93; I21; I25
Secondary IDs
Prior work
This trial does not extend or rely on any prior RCTs.
Abstract
Can responsible AI use make generative AI a complement to students’ skills rather than a substitute for skill-building effort? We study this question through a Responsible AI Use curriculum for Grade 6 and Grade 7 students. Forty schools are randomized to the curriculum or control. The curriculum teaches prompting, follow-up questioning, verification, critical judgment, and avoidance of answer copying. We distinguish AI-assisted performance from learning using standardized AI-use tasks and independently completed examination scores. We also test heterogeneity by baseline skills.
External Link(s)

Registration Citation

Citation
Meng, Xin et al. 2026. "What Skills Matter in the Age of AI? Evidence from a Responsible AI Curriculum in Rural China." AEA RCT Registry. September 21. https://doi.org/10.1257/rct.19607-1.0
Sponsors & Partners
Experimental Details

Interventions

Intervention(s)
The intervention is delivered by trained instructors and teaches students to use AI productively for learning rather than as an abstract technical subject. It focuses on problem decomposition, structured prompting, follow-up questioning, verification, creativity, critical judgment, and responsible use. Students use the Qwen web interface with pre-provisioned accounts. Each standardized 40-minute session combines a brief introduction, instructor demonstration, and guided practice.

The seven sessions cover an introduction to AI and responsible use, mathematics applications emphasizing structured problem solving and verification, Chinese applications emphasizing interpretation and textual evidence, and English conversation practice, with critical thinking and responsible AI use integrated throughout the curriculum.
Intervention Start Date
2026-09-22
Intervention End Date
2026-12-15

Primary Outcomes

Primary Outcomes (end points)
Administrative test score, performance in AI task, cognitive test
Primary Outcomes (explanation)
Administrative test score: Students’ scores on final examinations.
performance in AI task: Students complete standardized tasks using an AI chatbot. We record both task performance and detailed measures of AI use from the interaction logs.
cognitive test: Students complete a standardized 20-minute assessment of cognitive skills.

Secondary Outcomes

Secondary Outcomes (end points)
Additional secondary outcomes include survey items on creativity, critical thinking, self-regulation, mental health, curiosity.
For students in the treatment arm, Qwen interaction logs will provide complementary process measures of prompting, follow-up questioning, verification, response revision, and requests for direct answers.
Secondary Outcomes (explanation)

Experimental Design

Experimental Design
We randomize schools into treatment and control conditions using a stratified matched-pair design. Within each school level and urban/rural stratum, schools are ranked by the mean baseline Chinese and mathematics scores of students in the two preselected classes and paired adjacently. Within each pair, one school is randomly assigned to treatment and the other to control, yielding 20 schools per arm.

In treatment schools, the Responsible AI Use curriculum is delivered during regular ICT classes to the two selected classes. Control schools continue their regular ICT curriculum and receive no guided AI instruction. Students and instructors cannot be blinded to assignment, but routine examination grading will be blinded where feasible.
Experimental Design Details
Not available
Randomization Method
Randomization done in office by a computer
Randomization Unit
schools
Was the treatment clustered?
Yes

Experiment Characteristics

Sample size: planned number of clusters
40 schools
Sample size: planned number of observations
3200 pupils
Sample size (or number of clusters) by treatment arms
20 schools control, 20 schools treated
Minimum detectable effect size for main outcomes (accounting for sample design and clustering)
Assuming a school-level ICC of 0.10, baseline covariates explain 50 percent of outcome variance, a two-sided significance level of 0.1, and 80 percent power, the MDE is approximately 0.19 standard deviations. This calculation conservatively abstracts from additional precision gains from matched-pair randomization.
Supporting Documents and Materials
IRB

Institutional Review Boards (IRBs)

IRB Name
Research ethics committee of the China Center for Behavioral Economics and Finance
IRB Approval Date
2026-06-22
IRB Approval Number
2026SWB05
Analysis Plan