AI in Education & Learning Analytics

How India is deploying AI-powered adaptive learning across 1.5 million schools, reaching 260 million students — with RCT-proven results showing 2x learning gains in government schools at just $20 per student per year.

260M Students
182M DIKSHA Enrollments
₹500 Cr AI CoE Budget
#3 Global AI Rank

NEP 2020 — The Policy Backbone for AI in Education

India's National Education Policy 2020, replacing a 34-year-old framework, explicitly mandated AI, coding, and computational thinking at relevant stages of school curricula. The most consequential development came in October 2025, when the Ministry of Education announced AI & Computational Thinking as mandatory from Class 3 onwards, starting academic session 2026–27. An expert committee chaired by Professor Karthik Raman of IIT Madras developed the curriculum framework.

Scale of transformation: Over 7,500 PM SHRI schools upgraded as model NEP schools (target: 14,500 by 2027). CBSE offers AI as an elective for Classes 9–12 since 2019, and SOAR AI modules are active in 18,000+ schools. Budget 2025–26 allocated ₹1,28,650 crore to education with ₹500 crore earmarked for a dedicated Centre of Excellence in AI for Education.

Digital infrastructure is accelerating but uneven. UDISE+ 2024–25 data shows 63.5% of schools now have internet (up from 53.9%), 93.7% have electricity, and 64.7% have computers. But government schools lag at just 58.6% internet access versus 77.1% for private schools. States like Bihar (20.4% computer access), Assam (17.6%), and Madhya Pradesh (21%) remain severely underserved.

Government Platforms Operating at Massive Scale

DIKSHA
One Nation, One Digital Platform for Education

Launched in September 2017, DIKSHA operates across all 36 states and UTs in 36 Indian languages, hosting over 200,000 content pieces and 19,698 courses. Total enrollments have reached 182.3 million with 145.7 million completions and over 60 billion learning minutes logged. The NISHTHA teacher training program has issued 14 million certificates. A new AI feature, "AskDIKSHA," enables natural-language querying of NCERT books.

SWAYAM
India's National MOOC Platform

Covering Class 9 through postgraduate education, SWAYAM has offered over 14,829 courses through 203 partnering institutions, accumulating 56 million enrollments and issuing 5.2 million certificates. UGC now permits students to earn up to 40% of academic credits through SWAYAM courses.

PM eVidya
Multi-Modal Education Access

Expanded education through DIKSHA, 200 Swayam Prabha DTH TV channels, radio broadcasts, and content for children with disabilities including a dedicated Indian Sign Language channel. The program has generated 66,066 video programs and targets 250 million students.

The National Digital Education Architecture (NDEAR) provides the architectural blueprint, explicitly including "Open AI Services" — reusable AI libraries, models, and datasets. The IndiaAI Mission (₹10,372 crore over five years) includes FutureSkills for AI education and workforce skilling, with plans for Data and AI Labs in Tier 2 and Tier 3 cities.

State-by-State — RCT-Validated AI Deployments

Andhra Pradesh
Nobel Laureate-Backed RCT Proves 2x Learning Gains

The CG PAL (Personalized Adaptive Learning) program by ConveGenius.AI, evaluated by Nobel Laureate Michael Kremer's team at University of Chicago, produced the strongest evidence yet. In an RCT across 120 government schools with ~6,800 students over 17 months, treatment students achieved a 0.43 SD improvement — equivalent to 1.9 additional years of schooling in math. Students learned at 2.3x the rate of control peers, with girls gaining 2.31 equivalent years versus boys' 1.54. Cost: just ₹1,682 ($20) per student annually.

Separately, AP partnered with Microsoft to deploy Azure ML for dropout prediction, identifying 19,500 probable dropouts from Visakhapatnam district alone. The Vidyarthi Nestham App now operates in 10,000+ schools.

Kerala
First State to Add AI to State Board Curriculum

Kerala became the first Indian state to incorporate AI into its state board school curriculum in 2024, introducing a dedicated AI chapter for Class 7. KITE covers 16,000+ schools, 4.3 million students, and 170,000 teachers with 450,000+ ICT devices. Some 80,000 teachers received three-day AI training covering prompt engineering, ML, deepfakes, and algorithmic bias.

Karnataka
AI in 1,145 Schools + Face Recognition Attendance

The Kalika Deepa Programme with EkStep Foundation deploys AI-driven learning in math, English, and Kannada across 1,145 government schools reaching 144,000 students. The state also rolled out AI-based facial recognition attendance across ~50,000 institutions in October 2025, achieving 95% accuracy in pilot testing.

Tamil Nadu
22,931 Smart Boards in 7 Months

Installed 22,931 smart boards (₹455 crore investment) benefiting 1.18 million students. Establishing hi-tech labs with robotics and AI in 8,209 schools. The TN SPARK program delivers bilingual AI and robotics curriculum, while a Microsoft TEALS partnership covers 100 schools and 38,000 students.

Meghalaya
Pass Rates Jumped from 25% to 92%

iDream Education's deployment in 166 smart classrooms in East Jaintia Hills produced perhaps the most dramatic outcome: Class X pass rates improved from 25% to 92% within two years — a transformation in one of India's most remote and underserved regions.

More Notable Deployments

Uttar Pradesh's Mission Prerna reaches 18 million students across 160,000 schools. A NITI Aayog PAL pilot demonstrated 0.9 to 2.5 years of additional learning gains — India's first government-led outcome-based financing initiative. Rajasthan's J-PAL Mindspark RCT showed 0.2 SD improvement across 80 government schools. Maharashtra's "Hack the Classroom" pilot achieved a 40% reduction in grading time with 92% of students reporting greater confidence.

Gold-Standard RCT Evidence — AI Adaptive Learning Works

Study Platform Location Effect Key Finding
Muralidharan et al., 2019 (J-PAL) Mindspark Delhi, 5 schools 0.37 SD (math) 2–2.5x learning gains; first experimental CAL evidence
de Barros & Ganimian, 2023 (J-PAL) Mindspark Rajasthan, 15 schools 0.2 SD Weakest students benefit most from personalization
Kremer et al., 2025 (UChicago) CG PAL AP, 120 schools 0.43 SD 1.9 years additional learning; $20/student/year

The Delhi Mindspark study, published in the American Economic Review (2019), was the first experimental evaluation of adaptive learning software in a developing country to show clear results. At under $2/student/month, the intervention was 2–3x more cost-effective than comparable government school spending. An independent study estimated usage would increase lifetime earnings by 5–15%.

India's EdTech Ecosystem — AI at Extraordinary Scale

ConveGenius
WhatsApp-based SwiftChat in 13 languages; 150M+ student profiles; 10M daily messages; 98% completion rates
Embibe
104,464 schools in 25 states; 17M+ students; MoU with UP covering 55,000+ schools; 5 patents in personalized education
Physics Wallah
Alakh AI Suite (GPT-4o); 1.5M users in 2 months; AI Box transforms any TV into offline learning tool; 85% Tier 2/3 enrollment
Doubtnut
Image recognition for homework help in 11 languages; 32M monthly users; 90%+ from non-metro areas; acquired by Allen
Google Read Along
AI reading tutor; 7M+ students; 32M+ stories read; 64% improved reading proficiency in UP pilot across 200 villages
iDream Education
Offline-first iPrep PAL on tablets; 9 languages, 23 states, 1.5M+ students; Meghalaya 25%→92% pass rate transformation

The core technology stack across these platforms combines adaptive learning algorithms (personalized difficulty, knowledge graphs), NLP chatbots (WhatsApp-based doubt resolution), computer vision (OCR for homework scanning, facial recognition for attendance), predictive analytics (dropout prediction, learning gap identification), and speech recognition (reading proficiency assessment).

Challenges — Digital Divide, Teacher Readiness & Privacy

Over 500,000 schools still lack internet and 25,000+ lack working electricity. Urban tele-density stands at 134% versus rural at just 59%. BharatNet, designed to connect 600,000 villages, had reached only 214,000 by October 2024. Just 20% of Indians above age 5 possess basic digital literacy.

India needs to train 10 million+ teachers to deliver the new AI curriculum. An EY-FICCI survey found only 17% of university faculty consider themselves advanced AI users, and merely 6% are satisfied with institutional AI resources. The 2026–27 rollout timeline is extremely compressed.

Data privacy for minors is an emerging concern. The DPDP Act 2023 mandates verifiable parental consent for processing children's data and prohibits tracking, behavioral monitoring, and targeted advertising for children — with penalties up to ₹200 crore. A Human Rights Watch review found 89% of 163 global education apps collected children's data in ways risking their rights.

What This Means for the Future

India's AI-in-education journey offers critical insights. Adaptive learning technology works at scale — the AP RCT's 1.9 additional years of learning at $20/student/year makes it one of the most cost-effective development interventions available. WhatsApp-based delivery may be more impactful than app-based platforms in low-connectivity environments. India ranks 3rd globally on Stanford's AI Vibrancy Index — the only lower-middle-income country in the top three.

The October 2025 mandate making AI compulsory from Class 3 represents perhaps the largest structured AI curriculum integration globally by scale. The next 24 months will determine whether India's ambitions translate into measurable learning gains for the 130 million+ children still falling below grade-level competency.

Need AI Models or Software for Education?

AiBOUT builds custom AI solutions for educational institutions — from adaptive learning engines and student analytics dashboards to AI-powered attendance systems and automated assessment tools. Tell us what you need and we'll build it.

Contact Us

Research Sources

Back to Research