Improving reading and comprehension in K-12: Evidence from a large-scale AI technology intervention in India

Abstract

Abstract This paper presents evidence on the impact of technology-aided instruction on literacy using an AI-based multi-sensory technology platform across a large cross-section of government schools in India. The study focused on reading and comprehension in the English language. The intervention enhances the instructional effectiveness of the teachers and the learning ability of the children within the existing instructional environment without any new instructional design or pedagogy or content. Besides, the intervention is implemented by existing teachers and not outside volunteers. A total of 1 million children and 15,000 teachers across 5000 government schools in the states of Maharashtra, West Bengal, Punjab, Tamil Nadu, Telangana, Gujarat and Karnataka used the technology for the 2016-17 academic year. Using a randomized control-treatment assessment, the study finds a 20–40% overall gain in learning outcomes in the treatment sample. Gains within individual states and grades vary. Learning outcomes rose across the entire range of proficiency levels in a grade. Ongoing self-administered assessments report even higher impact in the 50–60% range. In addition, teachers also reported improving their skills as a result of using the technology, suggesting that the intervention can also alleviate teacher shortage and inadequate teacher training. The paper also reports briefly on the growing adoption of the program in several countries outside India. The results hold significant promise for disrupting the low and stagnating literacy levels across government schools in India and other similar environments.