BackHome
Refresh: OffView Personal Schedule
Session Submission Summary

AI-Enhanced Adaptive Teaching: From Evidence to Implementation

Mon, March 30, 4:30 to 5:45pm, Hilton, Floor: Ballroom Level - Tower 3, Continental 7

Group Submission Type: Formal Panel Session

Proposal

When learners fall behind academically, many disengage and drop out, creating cycles of exclusion that threaten social stability. This reality connects directly to CIES 2026's theme: persistent learning poverty maps onto social fault lines that can deepen division in fragile societies (World Bank, 2022). Adaptive teaching offers a pathway to interrupt these cycles by meeting learners at their current level and guiding them back into mainstream instruction.

This panel examines how artificial intelligence can strengthen adaptive teaching for foundational skills, particularly in classrooms with large student-teacher ratios and limited resources. We position AI not as teacher replacement but as practical support for generating level-appropriate tasks and enabling targeted remediation.

Structured pedagogy packages consistently produce significant gains in low- and middle-income systems, earning recognition as "good buys" for foundational literacy and numeracy (GEEAP, 2023). Teaching at the Right Level demonstrates that regrouping students by competency rather than age accelerates progress across diverse contexts (Banerjee et al., 2016). AI extends these approaches by making differentiation feasible in everyday classrooms through adaptive engines that calibrate difficulty and produce remedial activities aligned with curriculum standards. However, the Global Education Monitoring Report cautions that technology claims often exceed evidence, with LMIC-specific findings still too scarce for confident policy guidance (UNESCO, 2023). This panel therefore emphasizes empirical research testing when and for whom AI-supported adaptive teaching improves outcomes.

Recent studies indicate meaningful learning gains when personalization embeds within routine instruction. A randomized evaluation in India found students using personalized platforms improved by 0.37 standard deviations in mathematics and 0.23 in Hindi after 4.5 months, with largest gains among lower-performing learners (Muralidharan et al., 2019). This suggests adaptivity can narrow rather than widen gaps. Syntheses converge on key principles: digital practice works best when complementing teacher-led instruction, aligning with curricular sequence, and accompanied by teacher support rather than operating as parallel tracks (GEEAP, 2023).

Effective models must address infrastructure constraints in low- and middle-income contexts. Programs that succeed often operate offline with delayed sync capabilities and function on shared devices, recognizing connectivity and affordability challenges. Teacher agency remains critical, requiring approaches that support rather than substitute human interaction.

Research approaches in this field typically combine experimental designs with implementation studies to understand effectiveness across different contexts. Key questions include whether AI-supported adaptive teaching benefits different learner populations equitably and how to ensure responsible deployment that empowers rather than replaces educators.

This panel brings together four complementary perspectives examining both promise and limitation in AI-enhanced adaptive teaching. The contributions span from large-scale system implementation to localized pilots to global strategic analysis, providing the methodological diversity essential for advancing understanding in this emerging field.

The panel contributes to CIES 2026's theme by examining how AI might support educational equity and social cohesion. By bringing together diverse evidence types and implementation contexts, it advances understanding of both opportunities and risks. The session invites critical dialogue on when and how AI can support adaptive teaching without reinforcing inequities, encouraging participants to consider both its potential and its limitations.

Sub Unit

Organizer

Chair

Individual Presentations

Discussant