Fixing NEET starts with fixing the system
India has one of the largest examination ecosystems in the world. For many competitive examinations, the number of candidates exceeds the entire population of candidates in most foreign countries combined. So, we should appreciate that our challenge is at least an order of magnitude higher than in any other country.
These examinations determine admission not only to colleges and universities, but also entry into government and public sector jobs and many other careers. Their credibility is therefore fundamental to both individual opportunity and national confidence.
In the case of NEET in particular, the recurring cycle of question paper leaks, cancelled examinations, litigation, student anxiety, suicides, agitations and hunger strikes shows that this is no longer a mere administrative issue. It has become an educational, emotional, social and political challenge. Every breach weakens public trust in merit itself.
The real problem
The discussion after every leak usually revolves around identifying the guilty, increasing punishment, or tightening the confidentiality of the system. These measures have value, but they do not address the real problem.
After the next round of improvement, there is no guarantee that the next cycle of leak, investigation, punishment, change of minister, and one more round of improvement won’t repeat.
A secure examination system should not depend on assumptions about the personal integrity of any individual. Honest people certainly exist and should be respected. But institutions responsible for the future of millions must be designed on the assumption that at least one individual may eventually be compromised.
The system must remain secure even when individual integrity fails. Security must come from the architecture of the examination system, not from trust.
An independent examination authority
Establishing an independent constitutional National Examination Authority (NEA) is necessary but not sufficient. That it should function professionally and autonomously is a given. It can remain subject to the oversight of the Centre and the states, while staying insulated from day-to-day political and bureaucratic influence. Such an institution should provide continuity, technical expertise and public confidence across all major examinations.
What follows is an outline of an architecture that would be almost tamper-proof. Though the solution is presented here for NEET, it is equally applicable to any competitive examination, at the state or national level.
The architecture
The architecture itself is straightforward.
1. Crowdsourced question submission
The examination body should invite suggested questions and corresponding answers from bona fide teachers of the relevant subjects across India. This converts the country’s enormous teaching community into a continuous source of high-quality examination material, instead of relying on a handful of paper setters.
2. One-way, encrypted data flow
Every question submission should travel directly into highly distributed, encrypted databases, organised chapter-wise (and sub-topic-wise where needed), and tagged with syllabus codes. There should be no informal circulation, emailing or manual consolidation. The system should permit only one-way movement, from submission to validation and finally into the shortlisted question bank, never in reverse.
3. First-level AI screening
Multiple independent AI systems should perform only the first level of evaluation. Their task is to reduce volume, not replace academic judgement. They should check whether a question belongs to the prescribed syllabus, confirm the topic and sub-topic, detect duplicates, identify ambiguity, estimate quality, and flag likely errors in the proposed answer. Using multiple AI systems and agents, rather than one, reduces dependence on a single algorithm and improves robustness.
4. Independent human validation
Every shortlisted question should undergo human validation. A hundred or more competent paper setters across the country, unaware of one another, could be involved in this process. Each question should be independently evaluated by three teachers. Questions approved unanimously by all three enter the validated question bank immediately. If only two approve, the question should automatically be sent to two additional teachers, and accepted only if both of them approve as well. This substantially improves quality while eliminating dependence on any single individual.
The quality of question setting by these teachers should itself be continuously assessed by AI. Peer review of questions, as described above, can identify reviewers whose assessments fall below a high benchmark, so that such reviewers can be excluded from future validation exercises. The system thereby continuously improves itself.
5. Dynamic paper generation
The examination should not begin with a pre-assembled paper. Instead, the online system should dynamically generate each candidate’s question paper by pulling questions from multiple validated databases, according to predetermined chapter-wise proportions and difficulty bands. The exact timing of when questions are pulled can be determined by the NEA.
Every candidate would therefore receive a different, but academically near-equivalent, paper. Because there is no fixed paper waiting to be leaked, the opportunity for organised leakage is dramatically reduced. Even if some teachers were to leak questions they had submitted, there would be no guarantee that those particular questions had been accepted, let alone served to a particular student. At best, such a leak would expose a small subset of potential questions, which would mean little for the overall examination.
The architecture should permit only one-way movement of information: questions move from the database to the examination engine, never in the reverse direction.
A two-stage structure where needed
Most competitive examinations can stop at the objective stage itself. Only examinations that genuinely require higher-order reasoning or written analysis need a second, descriptive stage, limited to candidates who qualify in the first stage objective examination. This mirrors the structure of the IIT-JEE Advanced examination, and significantly reduces evaluation effort while preserving academic rigour where it is needed. A leak at this later stage is rare and far more easily controlled.
Fair scoring through normalisation
Fairness should continue even after the objective stage. Because questions are randomised at multiple levels, the probability is high that nearly all question papers will be of a similar level of difficulty. Even so, since every candidate receives a different combination of questions, each question can be statistically normalised based on the evaluation. The system should determine what percentage of candidates answered each question correctly, and assign a difficulty weight accordingly, so that harder questions carry slightly more weight than easier ones. Candidates should be rewarded for knowledge, not for the luck of drawing an easier set of questions, even if the probability of this may be negligible.
If taking the test on a later date offers any statistical advantage, this too can be normalised, based on the comparative performance of students across different dates.
Advantages of the architecture
Such an architecture offers multiple advantages. It nearly eliminates single points of failure, continuously improves question quality, scales naturally to millions of candidates, minimises not just the chances of fraud but also its incentives, allows continuous expansion of the question bank, improves transparency, and restores confidence that merit, rather than access, is deciding outcomes.
Beyond NEET
Most importantly, this proposal is not confined to NEET. The same architecture can be adopted for engineering entrance examinations, university admissions, recruitment examinations, public service commissions, civil services, banking recruitment, railways, defence recruitment, and every other large-scale competitive examination in the country.
Prevention by design
India needs a permanent institutional architecture designed to withstand human failure, technological threats and organisational pressure. The objective is not merely to prevent the next paper leak. It is to build an examination system worthy of the trust that millions of young Indians place in it every year, a system in which almost nothing can go wrong. And no more education ministers need to resign.
Real reform lies not in punishment after failure, but in prevention by design.
Note:
1. Text in Blue points to additional data on the topic.
2. The views expressed here are those of the author and do not necessarily represent or reflect the views of PGurus.
For all the latest updates, download PGurus App.
- NEET: Proposing a leak-proof architecture - August 10, 2026
- The regime change franchise: India is on the list too (A fictional thriller) - August 7, 2026
- Beyond NEET: Time to reform medical admissions - July 27, 2026
