Leveraging the Transformational Potential of AI to Empower Small-Scale Farmers in Ethiopia

How to cite

Miguel, Jérémy Do Nascimento; Abate, Gashaw; Bernard, Tanguy; Deutschmann, Joshua. 2026. "Leveraging the Transformational Potential of AI to Empower Small-Scale Farmers in Ethiopia." AEA RCT Registry. AEARCTR-0018829. https://doi.org/10.1257/rct.18829-1.0

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

Jérémy Do Nascimento Miguel* , Gashaw Abate , Tanguy Bernard , Joshua Deutschmann

* Primary investigator

Registry ID
AEARCTR-0018829
DOI
https://doi.org/10.1257/rct.18829-1.0
First registered
Last updated
Status
On going
Trial dates
2026-05-01 – 2028-01-01
Countries
Ethiopia (Oromia)
Registry record
http://www.socialscienceregistry.org/trials/18829

Abstract

This project evaluates whether artificial intelligence (AI) can help improve agricultural advice and decision-making among smallholder farmers in Ethiopia. The study focuses on FarmerChat, an AI-powered mobile application that allows farmers and agricultural extension agents to ask questions using text, voice, or pictures and receive tailored recommendations on crop production, livestock management, weather conditions, and agricultural practices in local languages.

The project will be implemented in 317 villages across 15 districts of the Oromia region and will involve more than 2,600 farming households and over 300 agricultural extension agents. Villages will be randomly assigned to one of three groups: a control group, a group where extension agents receive training on FarmerChat, and a group where extension agents are trained and encouraged to actively introduce the application to farmers. This randomized design will allow us to rigorously measure the impact of AI-based agricultural advice.

Using household surveys and administrative data from the application, we will examine whether access to FarmerChat increases engagement with digital advisory services, improves agricultural knowledge and confidence, encourages the adoption of improved farming practices, and ultimately increases agricultural productivity and incomes. The study will also assess whether AI-powered advisory services can help reach farmers who are traditionally underserved by extension systems, including women and farmers living in remote areas.

Intervention

The project will be implemented in 317 villages across 15 districts of the Oromia region and will involve more than 2,600 farming households and over 300 agricultural extension agents. Villages will be randomly assigned to one of three groups: a control group, a group where extension agents receive training on FarmerChat, and a group where extension agents are trained and encouraged to actively introduce the application to farmers. This randomized design will allow us to rigorously measure the impact of AI-based agricultural advice.

Experimental design

The study uses a cluster-randomized controlled trial conducted in 317 kebeles (villages) across 15 woredas in the Oromia region of Ethiopia. Kebeles are randomly assigned to one of three groups: (i) a control group receiving no FarmerChat onboarding, (ii) a treatment group in which Development Agents (DAs) receive training and onboarding to the FarmerChat platform, and (iii) a treatment group in which DAs receive the same onboarding and are additionally encouraged to actively onboard farmers with mobile phones in selected villages. Randomization is stratified by woreda and implemented at the kebele level to minimize spillovers.

Design details

Randomization method
Randomization done in office by a computer
Randomization unit
Kebele: smallest administrative units in Ethiopia
Clusters
317 kebeles
Observations
2,600 farmers and 317 development agents
Arms
105 kebele control, 106 kebele DAs only treatment, 106 kebele DAs + farmers onboarding treatment
Primary outcomes
The primary outcomes of interest are engagement with FarmerChat, agricultural awareness and knowledge, confidence in agricultural decision-making, adoption of improved agricultural practices, and agricultural productivity.
Secondary outcomes
Secondary outcomes include perceptions of FarmerChat, food security, food consumption, technology-related anxiety, aspirations, and other behavioral measures.
Partners
Name: Digital Green Partner_type: ngo Url: https://digitalgreen.org

Cited by

No citation data available for this registration yet.

Keywords