Why in News?
· India's test-preparation market is estimated at $14.8 billion in FY2026, projected to reach $23-26 billion by FY2030.
· The 2025 NSO survey indicates that around 27% of students take or have taken private coaching, reflecting the dependence on paid supplementary education.
· ASER 2024 found that only 45.8% of Class VIII children could perform basic division, showing that years of schooling do not automatically ensure learning.
· India has around 24.8 crore school students, 14.72 lakh schools and 98 lakh teachers, making individualised conventional tutoring difficult at national scale.
· AI can provide personalised diagnosis, instant feedback, adaptive practice and 24×7 doubt-solving at relatively low marginal cost.
Need for an AI-enabled Education Ecosystem
Persistent learning gaps: AI can identify the precise source of conceptual weakness rather than merely reporting marks.
- Example: A student's algebraic errors can be traced to weak fractions or arithmetic.
Coaching affordability: High-cost coaching creates an income-linked advantage in competitive examinations.
- Example: Free AI tutoring can provide academic support to a government-school student unable to afford private coaching.
Content abundance: India has extensive digital educational material, but students face problems of discoverability, sequencing and personalisation.
- Example: A NEET aspirant may find thousands of biology lectures but not know which prerequisite concepts to study first.
Existing digital foundation: DIKSHA and NDEAR, supported by NEP 2020, provide foundations for interoperable digital education.
Personalised pedagogy: AI can shift education from one-size-fits-all instruction to competency-based learning.
- Example: Two students studying calculus can receive different practice sets according to their individual error patterns.
Public Rails, Private Engines and Unbundling Coaching
Government as infrastructure provider: The state should provide authentication, standards, APIs, accreditation and learning records, rather than becoming the country's largest coaching provider.
Private-sector participation: Private firms, teachers and universities can supply lectures, question banks, mentoring, mock tests and specialised tutoring.
DPI precedent: UPI demonstrates how common public infrastructure can enable competition among multiple private providers.
Open educational network: An open network can prevent students from being locked into one coaching platform.
- Example: A student could use mathematics content from one provider and physics content from another.
Demand aggregation: Government can aggregate millions of students, reducing customer-acquisition costs and encouraging providers to compete on quality.
Architecture and Functioning of the AI Tutor
Content registry: A common registry should classify resources by subject, topic, difficulty, language, examination and learning outcome.
- Example: “NEET Biology–Genetics–Mendelian inheritance” becomes a common learning node.
Portable learning record: Students should retain lessons completed, mock scores and topic mastery when changing providers.
- Example: Switching coaching platforms should not erase a student's learning history.
Diagnostic engine: AI can identify specific conceptual weaknesses.
- Example: Repeated calculus errors can trigger remedial modules on functions and algebra.
Adaptive tutor: AI can progressively change question difficulty according to performance.
- Example: Weak performance triggers foundational questions before examination-level problems.
Learning wallet: An optional e-Shiksha wallet could finance human mentoring, graded assignments or specialised test series for disadvantaged students.
Equity, Inclusion and Human-AI Learning
Digital divide: ASER 2024 found that about 90% of 14-16-year-olds reported smartphone access at home, but personal ownership was substantially lower, with a gender gap.
Infrastructure gap: Schools with computers increased from 38.5% in 2019-20 to 57.2% in 2023-24, while internet access increased from 22.3% to 53.9%.
Rural accessibility: Offline content, low-bandwidth applications and Common Service Centres can reach students with weak connectivity.
- Example: Lessons can be downloaded and AI-assisted practice synchronised when connectivity returns.
Multilingual education: AI can explain concepts in regional languages while retaining examination terminology in English.
- Example: A student can learn a physics concept in their mother tongue and practise English technical vocabulary.
Human-AI complementarity: AI can handle repetitive tutoring, while teachers provide motivation, advanced guidance and socio-emotional support.
- UNESCO's guidance on AI in education emphasises human-centred and pedagogically appropriate AI deployment.
Implementation Challenges
AI accuracy and hallucinations: Incorrect explanations can create large-scale learning losses.
- Example: An erroneous chemistry explanation could be repeatedly provided to thousands of students.
Digital inequality: Unequal access to devices, electricity and connectivity could reproduce existing educational inequalities.
- Example: A free AI tutor remains inaccessible to a student without a smartphone or nearby digital centre.
Data privacy: AI systems can collect detailed learning profiles containing academic weaknesses, interests and behavioural patterns.
- Example: Student learning data could be misused for targeted advertising or commercial profiling.
Commercial concentration: Large ed-tech firms may dominate the supposedly open ecosystem.
- Example: Recommendation algorithms could systematically prioritise commercially powerful providers.
Teacher displacement and overdependence: Excessive reliance on AI could weaken human interaction, mentoring and classroom learning.
- UNESCO stresses human agency and warns against treating technology as a substitute for educational institutions and teachers.
Academic integrity and bias: AI-generated answers can facilitate cheating, while algorithms may favour particular languages or providers.
- Example: An AI recommendation system trained predominantly on English content could disadvantage students using regional languages.
Way Forward
Establish Education DPI governance: Create an independent Education DPI steward responsible for interoperability, accreditation, standards and grievance redressal.
- Example: A common protocol can allow DIKSHA, State platforms and private providers to exchange learning resources.
Begin with JEE and NEET: Pilot the system for examinations with standardised syllabi, objective assessment and measurable outcomes before expanding to CUET and State examinations.
Build verified AI tutors: Develop curriculum-grounded systems using expert-validated content, continuous benchmarking and human escalation.
- Example: AI should transfer an uncertain advanced physics problem to a qualified teacher rather than generate a speculative answer.
Guarantee universal access: Expand school computer laboratories, CSC access, downloadable content, offline functionality and low-bandwidth services.
- Example: Rural students can access lessons offline and synchronise their learning records later.
Protect student data: Ensure data minimisation, consent-based sharing, portability and student control over learning records.
- Example: Students should retain their learning history when moving between providers.
Create outcome-based accountability: Publish transparent indicators covering learning gains, standardised mock performance, completion and examination outcomes.
- Example: Providers should compete on measurable improvement rather than merely advertising rank-holders.
Strengthen teachers and AI literacy: Train teachers to use AI for diagnosis and personalised instruction while teaching students AI literacy, critical thinking and verification skills.
- Example: Teachers can use AI-generated class-level diagnostics but retain responsibility for instructional decisions.
