VTU to Use AI for Question Papers: What India's Universities Are Watching

A Major University Takes AI Beyond the Classroom

Visvesvaraya Technological University (VTU) in Belagavi, Karnataka — one of India's largest affiliating technical universities with over 200 affiliated colleges — has announced that it will use artificial intelligence to generate question papers, beginning with a pilot on MBA examinations.

Vice Chancellor Vidyashankar S told Deccan Herald that the university intends to start with MBA because the student numbers are smaller than engineering, making it a more manageable proving ground. Subject experts will cross-verify each AI-generated paper before it goes to print. Once the MBA pilot demonstrates results, the model is expected to extend to the full range of B.E. programmes. The VC also said he wants a second evaluation of answer sheets to be done using AI, though no specific timeline has been given for that part.

This announcement places VTU among a small but growing group of Indian universities explicitly committing to AI as a tool within their core examination function — not just in teaching, research, or administrative automation.

Why Question Paper Quality Has Long Been a Problem

The difficulty of setting high-quality, syllabus-aligned question papers at scale is underappreciated by those outside examination administration. A large affiliating university must manage question papers across hundreds of subjects, multiple programmes, and several examination cycles every year. The pool of available paper-setters is often limited. Repeat questions, out-of-syllabus questions, and papers with missing sections appear regularly enough to have their own category in student grievance records.

VTU's VC has acknowledged this directly, saying the current standard of question papers needs improvement. This is an unusual admission from a university leadership, and it sets up the AI pilot as a corrective measure rather than a purely futuristic experiment.

AI-based paper generation works by ingesting syllabus content, bloom's taxonomy levels, past paper patterns, and difficulty-distribution requirements, then producing question sets that meet those parameters. Subject expert review is the essential human check — it catches hallucinations, confirms technical accuracy, and ensures that the output meets discipline-specific norms that AI does not reliably infer.

How the Pilot Is Expected to Work

Based on the VTU VC's stated approach, the process will likely follow a pattern similar to what other institutions have piloted:

The company for the AI tool had not been chosen at the time of the announcement. VTU does not appear to be accelerating into an untested rollout; the phased approach with MBA first mirrors how the university introduced other examination technology changes in recent years.

The Broader AI-in-Examination Trend

VTU's announcement joins a cluster of Indian university moves in the same direction.

At AI Manthan 2026, a January workshop jointly organized by AKTU and CSJMU Kanpur, participants discussed AI-based examination systems and automated question paper generation at length. The CSJMU vice-chancellor publicly discussed plans for AI to assist in paper setting. In Karnataka, Deccan Herald reported separately that VTU has already introduced AI across every engineering curriculum, which gives it a faculty base familiar with the tools being applied to paper generation.

In Tamil Nadu and Telangana, state university officials have discussed AI-assisted evaluation for subjective papers, though firm rollout announcements have been more cautious. Osmania University in Hyderabad and JNTU-H have been running on-screen marking for several years, providing the scanning and evaluation infrastructure that an AI-assisted second valuation would require.

What the VC Said About AI Evaluation of Answer Sheets

Beyond question papers, the VTU announcement carries a second, larger implication. The VC explicitly said the goal is to have a second evaluation of answer sheets done by AI. This would represent a meaningful shift in how subjective university exams are checked.

Currently, most Indian universities run double valuation — where a second human evaluator independently marks scripts that fall within a certain range of the first evaluator's marks. AI second valuation would replace one human evaluator with an AI system for this step. The stated benefits include speed, consistency, and removing the human evaluator availability bottleneck that causes re-evaluation backlogs at the end of every semester.

Whether AI subjective evaluation can reliably match human judgment at the level required for high-stakes university examinations remains contested. The Jodhpur district school pilot — where AI evaluated over 70,000 student papers in school subjects — showed speed and cost advantages, but involved structured, shorter-format responses different from university-level B.E. and MBA papers.

Risks and Considerations for Institutions Watching This Space

VTU's pilot will be watched closely by university examination controllers across India. Several questions matter:

Accuracy and appeals: If AI-generated papers contain errors or if AI-evaluated answer sheets produce disputed marks, what are the redressal mechanisms? Universities need clear policy before deploying AI evaluation at scale.

Vendor lock-in: VTU has not selected its AI vendor. The company and the contractual terms around data custody, model auditing, and system continuity will affect every downstream question of accountability.

Regulatory standing: UGC and AICTE have not yet issued guidelines on AI-generated question papers or AI-assisted evaluation for affiliating universities. Institutions piloting ahead of formal guidance take on regulatory ambiguity.

Staff training: For AI paper generation to work well, subject faculty need training on the prompt structures and quality-check criteria for reviewing AI outputs. This is a curriculum design skill, not a technology skill.

What This Means for the Examination Infrastructure Stack

Universities considering any part of this trajectory — AI paper generation, AI evaluation, or simply modernizing their examination workflow — face the same underlying requirement: a robust digital examination infrastructure must be in place first.

Scanning answer sheets reliably, maintaining evaluator login trails, capturing page-level marks, enabling remote evaluation, and preserving a full audit record are preconditions for any AI layer on top. Universities that have not yet digitized their answer sheet handling will find that AI paper setting is the easy part; AI-assisted evaluation requires far more of the upstream infrastructure to be in place before it adds value.

VTU's 5-minute result turnaround, already reported earlier this year, suggests its back-end systems are more developed than many affiliating universities. The MBA AI paper pilot sits atop that base.

Related Reading