← Research

AI for Education

Using large language models to build better educational tools: automated question generation aligned to Bloom’s skill levels for Indian high-school curricula, and EvalYaks — instruction-tuning datasets and LoRA fine-tuned models for automated scoring of CEFR B2 speaking assessments. An ongoing PhD is extending this into AI-enabled personalized adaptive learning systems, and this pillar is one of many touched by the lab’s emerging Agentic AI direction, which now spans all seven research pillars.

Publications in this pillar

1. Learning in Blocks: A Multi Agent Debate Assisted Personalized Adaptive Learning Framework for Language Learning

   Nicy Scaria, Silvester John Joseph Kennedy, Deepak Subramani · International Conference on Artificial Intelligence in Education (AIED) 2026, pp. 399-413 · 2026

2. EvalYaks: Instruction tuning datasets and LoRA fine-tuned models for automated scoring of CEFR B2 speaking assessment transcripts

   Scaria, N., Kennedy, S. J. J., Latinovich, T., Subramani, D. · Computers and Education: Artificial Intelligence, 100539 · 2025

3. Automated Educational Question Generation at Different Bloom's Skill Levels Using Large Language Models: Strategies and Evaluation

   Scaria, N., Dharani Chenna, S., Subramani, D. · AIED 2024. Lecture Notes in Computer Science, vol 14830. Springer, Cham · 2024 ★ CORE2023: A Ranking, Accepted as Main Paper (Oral). Acceptance rate = 23%

4. How Good are Modern LLMs in Generating Relevant and High-Quality Questions at Different Bloom's Skill Levels for Indian High School Social Science Curriculum?

   Nicy Scaria, Suma Dharani Chenna, Deepak Subramani · BEA 2024, pages 1–10, Mexico City, Mexico. ACL · 2024

All AI for Education publications on the Publications page →