AI-assisted teaching in Pakistan

Last registered on November 10, 2025
View Trial History

Pre-Trial

Trial Information

General Information

Title
AI-assisted teaching in Pakistan
RCT ID
AEARCTR-0017128
Initial registration date
November 07, 2025

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
November 10, 2025, 10:06 AM EST

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
Jonah Rexer
Affiliation
The World Bank
Contact Primary Investigator

Other Primary Investigator(s)

PI Name
Siddharth Sharma
PI Affiliation
The World Bank
Contact Investigator
PI Name
Jishnu Das
PI Affiliation
Georgetown University
Contact Investigator
PI Name
Saher Asad
PI Affiliation
The World Bank
Contact Investigator

Additional Trial Information

Status
In development
Start date
2025-12-01
End date
2026-12-31
Keywords
Education
Additional Keywords
JEL code(s)
Secondary IDs
Prior work
This trial does not extend or rely on any prior RCTs.
Abstract
This study investigates whether artificial intelligence (AI) can improve learning outcomes in Pakistan. Foundational literacy and numeracy (FLN) remain critical challenges in the country; according to the 2023 Annual Status of Education Report, only 50 percent of grade 5 in Pakistan students can read a simple story in Urdu, Sindhi, or Pashto, and just 46 percent can do two-digit division (ITA 2024). In addition, there is growing evidence pointing to significant learning losses experienced by students during breaks between school semesters and academic years. In particular, recent studies show that children in low- and middle-income countries often lose the equivalent of several months of learning during school closures or long breaks, with the youngest and most disadvantaged students falling furthest behind (World Bank, 2022). In response to this challenge, the study explores whether a structured winter school program can mitigate these learning losses, and whether incorporating AI-based tools into teaching practices can enhance its effectiveness.

The intervention involves a winter school program centered on the Targeted Instruction Program (TIP) curriculum. The TIP curriculum focuses on foundational competencies in Urdu, mathematics, and English. The winter school program will serve students in grades 2 through 5 in Quetta, Balochistan, Pakistan during the extended winter break period. Balochistan has particularly poor academic outcomes, with just 46 and 26 percent of students able to read a simple story in their native language or do two-digit division, respectively. In addition to the core academic instruction, the intervention includes teacher training and the integration of AI-assisted lesson planning, grading, and educational material support.
External Link(s)

Registration Citation

Citation
Asad, Saher et al. 2025. "AI-assisted teaching in Pakistan." AEA RCT Registry. November 10. https://doi.org/10.1257/rct.17128-1.0
Sponsors & Partners

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

Request Information
Experimental Details

Interventions

Intervention(s)
The intervention involves a winter school program centered on the Targeted Instruction Program (TIP) curriculum. The TIP curriculum focuses on foundational competencies in Urdu, mathematics, and English. The winter school program will serve students in grades 2 through 5 in Quetta, Balochistan, Pakistan during the extended winter break period. Balochistan has particularly poor academic outcomes, with just 46 and 26 percent of students able to read a simple story in their native language or do two-digit division, respectively. In addition to the core academic instruction, the intervention includes teacher training and the integration of AI-assisted lesson planning, grading, and educational material support.
Intervention Start Date
2026-01-01
Intervention End Date
2026-03-01

Primary Outcomes

Primary Outcomes (end points)
The primary outcome of interest is a composite test score combining performance in Urdu, math, and English. A composite measure is used to maintain statistical power and avoid the need for multiple hypothesis correction. All 9,000 students who sign up will complete a baseline assessment conducted in their home schools. After the winter school program concludes, endline testing will be conducted. A follow-up assessment six months to one year after the intervention will evaluate the persistence of learning gains. Winter school teachers will also be surveyed on their background, characteristics, and time use, at baseline and endline. Basic demographic data will be collected from all students using take-home forms and existing student registries at the time of the baseline survey. Prior test scores if available will be collected at origin schools.

Additional data will be collected to monitor program implementation and compliance. Metadata from the AI tool will be leveraged to measure engagement. Teacher and student attendance, as well as tool usage, will be recorded as part of the treatment data, providing outcomes for a “first-stage” measuring take-up of both the winter school and the AI technology. Compliance will be monitored through AI-assisted image checks and occasional spot checks or site visits.
Primary Outcomes (explanation)

Secondary Outcomes

Secondary Outcomes (end points)
Secondary Outcomes (explanation)

Experimental Design

Experimental Design
The study will employ a randomized controlled trial (RCT) with three treatment arms

Control: a control group receiving no intervention,

T1: a group attending the winter school without AI.

T2: a group attending the winter school with AI-supported instruction.

Students will be recruited from their existing schools, and participation will be voluntary. All students who sign up will be tested at baseline. These students will then be randomly assigned to one of the three groups, ensuring student-level randomization. Teachers will also voluntarily sign up to teach in the winter school and will be randomly assigned to classrooms either with or without AI assistance. To encourage participation, students will receive transportation and meal allowances, while teachers will be compensated with additional salary. The winter school will follow a student-teacher ratio of 40:1.
Experimental Design Details
Not available
Randomization Method
Public lottery
Randomization Unit
Child-level randomization
Was the treatment clustered?
No

Experiment Characteristics

Sample size: planned number of clusters
9000 students
Sample size: planned number of observations
9000 students
Sample size (or number of clusters) by treatment arms
3000 students per treatment arm (T1, T2, and Control)
Minimum detectable effect size for main outcomes (accounting for sample design and clustering)
The study will be powered to detect a minimum detectable effect (MDE) of 0.1 standard deviations between any two groups. This means that in order to detect the differences between both T1 vs. C and T2 vs. T1, T1 must deliver a 0.1SD increase from C, while T2 must deliver a 0.2SD increase from C.
Supporting Documents and Materials
IRB

Institutional Review Boards (IRBs)

IRB Name
HML IRB
IRB Approval Date
2025-11-06
IRB Approval Number
3166
Analysis Plan