Early Exposure to AI Training and Students’ Educational Trajectories in Ethiopia

This pilot evaluates the impact of early exposure to AI-powered learning tools combined with teacher training on educational outcomes in Ethiopia. The intervention has two components: (i) student access to adaptive, SMS-based AI learning platforms that provide personalized lessons, assessments, feedback, and AI literacy modules aligned with the national curriculum; and (ii) teacher professional development, including workshops and ongoing support to integrate AI tools into classroom instruction using data-driven and differentiated teaching practices. Students in the treatment group will receive both components, while the control group continues with standard curriculum and teaching methods without AI support.

2026-09-07

2026-12-31

Students’ academic performance (math and English test scores), student engagement, digital readiness, interest in STEM pathways, and teacher retention.

Academic performance will be measured using standardized math and English test scores at baseline and endline. Student engagement will be captured through attendance records and platform usage data (e.g., lesson completion and frequency of interaction). Teacher retention will be measured using administrative records and surveys on turnover and job satisfaction. Digital readiness will be assessed through task-based evaluations of students’ and teachers’ ability to use digital tools. Interest in STEM pathways will be measured through surveys capturing students’ subject preferences and future educational aspirations.

Students’ digital literacy skills, teacher instructional practices (effectiveness), students’ educational aspirations, and classroom learning environment quality.

Students’ digital literacy will be measured through task-based assessments of their ability to use digital tools and platforms. Teacher instructional practices will be evaluated using classroom observations and structured surveys capturing the use of differentiated instruction and data-driven teaching. Students’ educational aspirations will be measured through survey questions on expected years of schooling and intended career paths. Classroom learning environment quality will be assessed using student surveys and observation tools capturing engagement, inclusiveness, and responsiveness to diverse learning needs.

Randomized Controlled Trial (RCT).

Not available

RCT design with individual-level random assignment, stratified by gender and grade level.

Individual (student)

No

N/A

300 Individual (student). Approximately equal allocation between treatment (T1) and control (T0).

N/A

Approximately 0.20–0.25 standard deviations for key academic outcomes, assuming 80% power and a 5% significance level.