AIED Unplugged: Expanding Learning Opportunities and Reducing Educational Inequities in Low- and Middle-Income Countries

• In Event: AI-Enhanced Adaptive Teaching: From Evidence to Implementation

Proposal

Educational inequities remain a pressing challenge across the Global South, where limited internet connectivity, fragile digital infrastructure, and low levels of digital literacy prevent millions of students from accessing technology's potential in education. This gap is further widened by the "AIED Divide" — the growing disparity in the capacity to adopt and benefit from Artificial Intelligence in Education (AIED). Without targeted innovation, AI risks reinforcing existing inequalities.
AIED Unplugged is a scalable, inclusive framework designed for low-resource environments, enabling AI-supported learning without the need for constant internet access, personal devices, or advanced digital skills. Rooted in public policy and designed to integrate with existing school infrastructure, the approach has been implemented in Brazil, Chile and is now expanding to El Salvador. Across three large-scale applications — foundational literacy (over 8,000 schools and 164,000+ students in Brazil), numeracy skills (offline AI analysis of 5,200+ handwritten exercises), and dropout risk prediction (reaching 16,470 students in Brazil and informing Chilean national policy) — AIED Unplugged has demonstrated measurable gains in learning outcomes and reductions in educational disparities.
This work directly contributes to the panel theme "AI for Tech-Enhanced Adaptive Teaching", which will explore how artificial intelligence can generate personalized support for teachers and struggling learners to address persistent learning inequities that often contribute to broader social divisions. This panel presentation will discuss the policy design, implementation strategies, and impact evidence of AIED Unplugged, highlighting its potential to expand equitable learning opportunities across low- and middle-income countries.

Author

• Ig Ibert Bittencourt, FUNDACAO UNIVERSITARIA DE DESENV DE EXTENSAO E PESQUISA