Imagine spending two years preparing for a competitive examination. You sacrifice sleep, money, family time and perhaps even other career opportunities. You enter the examination hall believing that your preparation will determine your result.
Then you hear that the question paper may have leaked.
Suddenly, the problem is no longer how well you studied. It is whether the system gave you a fair chance in the first place.
This is the crisis that India's proposed examination reforms must address. And now, technology is being presented as part of the answer.
Infosys co-founder Nandan Nilekani, who heads the Centre's High Powered Task Force on examination reforms, has outlined an approach that could change how question papers are prepared and delivered. The proposal involves a secure digital question vault containing more than one lakh vetted and translated questions, from which papers could be assembled shortly before an examination.
The idea sounds promising. But an important question remains: Will it make examinations genuinely more secure, or will it simply move the vulnerabilities somewhere else?
Not sharing a Solution but A Way to shift the Accountability -> AI/COMPUTER
The logic behind a question vault is straightforward.
Instead of preparing one fixed question paper that must be protected until examination day, authorities would maintain a large bank of questions. A specific paper would be assembled by selecting questions from that bank shortly before the test.
If implemented properly, this could reduce the time during which a complete examination paper exists in a form that someone might steal or sell.
It could also make it harder for insiders to predict exactly which questions will appear.
That is a meaningful improvement in principle.
But there is a difference between making a leak more difficult and making an examination system secure.
A question does not enter a digital vault by magic. Someone must write it, review it, translate it, approve it and upload it. Software must select it, systems must deliver it, and examination centres must administer the test.
Every stage creates its own requirements for security and accountability, and no one here who wants to take accountability.
The SSC example offers hope, but it deserves scrutiny
India's (SSC) Staff Selection Commission already uses a version of this model for its computer-based recruitment examinations.
According to SSC Chairman S. Gopalakrishnan, questions are uploaded by subject experts, vetted, verified and translated before being stored in an internal digital vault. Shortly before an examination, questions are randomly selected according to a predetermined design and blueprint.
The paper is then compiled, encrypted and delivered to examination centres. Candidates can access it only at the scheduled time after logging in.
Gopalakrishnan has said that the SSC has not recorded a single paper leak in the last five years.
If independently substantiated, that record would be encouraging. It suggests that digital question management and tightly controlled delivery may offer practical advantages over relying on a single pre-prepared paper.
However, one important distinction remains: a reported absence of recorded leaks does not automatically establish that every part of the system is secure, or that the same approach will work equally well for every examination.
Different examinations have ->
Different numbers of candidates, delivery arrangements, language requirements and security risks.
A question vault cannot automatically guarantee a fair examination
Consider a problem that receives less attention than paper leaks: fairness.
If "A" candidate receives a paper containing relatively straightforward questions, while another "B" receives a significantly more difficult combination.
Both papers may have been generated securely. Neither may have leaked. Yet the candidates may not have faced an equivalent test of their knowledge.
Random selection alone does not guarantee equal difficulty.
A question bank must be carefully designed, with questions tested and classified according to relevant factors such as subject coverage, difficulty and the skills being assessed. Different versions of an examination must be sufficiently comparable for results to be meaningful.
Translation adds another layer of complexity. A question that is clear in English may become ambiguous in another language. Even a small change in wording can alter what a question asks or how difficult it is to interpret.
This is especially important in a country where candidates may take examinations in different languages.
"Security is essential, but a perfectly protected paper that measures candidates unfairly is not a successful examination."
What happens if the weakness is inside the "System"?
There is another problem that technology alone cannot resolve.
A digital vault can restrict access, encrypt information and record activity. These safeguards can make unauthorized access harder and investigations more effective.
But they depend on how the system is designed, implemented and monitored.
- Who is allowed to upload questions?
- Can one person change a question after approval?
- How are access permissions reviewed?
- Are all important actions recorded?
- Who investigates suspicious activity?
- Can independent auditors examine the security arrangements?
- And what happens if a breach is discovered?
These are not arguments against technology. They are the questions that responsible technology adoption must answer.
The task force is also considering separating responsibilities such as question-paper preparation, examination-centre management and the conduct of tests. This is potentially important because concentrating too much control in one organisation can increase risk.
Nandan Nilekani pitches for AI generated question papers
Nilekani is giving "Lollipop" in the Name of "AI"?
Nilekani has also indicated that artificial intelligence could be used to generate examination questions on the day of the test. May be he don't know how AI work or he is just fooling People.
That possibility deserves a separate discussion.
Selecting vetted questions from an established bank is not the same as asking an AI system to create new questions shortly before an examination.
AI might eventually help with question generation, translation, classification and quality checks. But generated questions must still be evaluated for accuracy, ambiguity, bias, difficulty and alignment with the examination's syllabus.
A question can sound perfectly reasonable and still contain a factual error. A translation can be grammatically correct yet change the meaning. A generated set of questions can appear balanced without actually testing candidates at comparable levels.
In a high-stakes examination, these are not minor inconveniences. They can affect admissions, employment and a candidate's future.
"Before expanding its role, authorities should establish how the questions will be tested, who will approve them and how errors will be corrected."
Are we solving the problem, or just impressed by "TECHNOLOGICAL SOLUTION"?
This is the question worth asking, although it should not be confused with an accusation that the proposal is a commercial scheme.
The available report does not establish that the reform is designed to sell products or benefit a particular company. Claims about improper commercial motives would require evidence, such as procurement records, conflicts of interest or documented financial relationships.
Nevertheless, public scrutiny is entirely justified.
Whenever governments introduce a major technological system, citizens should be able to understand what it is intended to achieve, how much it costs, who is responsible for operating it and how its performance will be assessed.A secure digital platform may be necessary. But technology procurement is not the same as reform.
The public deserves to know whether the new system reduces risks in practice, whether its costs are justified, and whether independent experts can examine its security and fairness. It also deserves clear accountability when failures occur.
Otherwise, there is a danger that a technically impressive solution will become a substitute for the less glamorous work of improving oversight.
What would success actually look like?
The success of this proposal should not be measured by the number of questions stored in a vault, the sophistication of the software or the presence of AI in the process.
It should be measured by outcomes.
- Are examination leaks becoming less frequent?
- Are different versions of a test demonstrably comparable in difficulty?
- Are translations independently checked?
- Are access logs audited?
- Are breaches investigated transparently?
- Are candidates treated fairly when an examination is compromised?
The authorities should establish clear security standards, independent audits, transparent reporting and a credible process for handling suspected leaks. Candidates should also have access to a fair grievance and remedy mechanism when serious failures occur.
Technology may help protect the question paper. Only a well-designed, independently accountable system can give candidates reason to trust the examination.
What should be "Next topic"?