IIM Nagpur's AI Grading Pilot: From Two Weeks to 48 Hours

India's First Major AI Grading Experiment in Management Education

When the Indian Institute of Management Nagpur announced that it would use artificial intelligence to evaluate student answer sheets and project submissions, the education sector took notice. The stated goal: cut the standard two-week manual grading cycle down to 24 to 48 hours. If the pilot delivers, it would represent one of the most significant shifts in examination assessment in Indian higher education in decades.

IIM Nagpur's initiative is notable not just for its ambition but for its scope. The institution is exploring AI not only for evaluating written answer scripts, but also for setting question papers, with difficulty calibrated through structured AI prompts. In a further step, students at the institution are being assessed on their ability to use AI tools effectively — making the technology both the medium and the subject of evaluation.

What the System Actually Does

The AI grading system being explored at IIM Nagpur is not simply a spell-checker or an OMR scanner. It is an AI engine capable of reading handwritten or typed subjective answers, comparing them against model responses and rubrics, and assigning marks with explanations. The director of IIM Nagpur has described the rationale simply: manual assessment, which currently takes about two weeks, can be completed in a day or two through AI tools.

This is meaningfully different from what most Indian boards call "digital evaluation." The CBSE On-Screen Marking system, for instance, scans answer sheets and displays them on a screen for a human teacher to mark. The intelligence is human; the digital layer is logistical. IIM Nagpur is exploring a system where the AI itself reads, interprets, and scores the answer — a genuinely different proposition.

Broader research supports the efficiency case. Studies on AI-assisted grading have found reductions of approximately 31 percent in time per response and 33 percent per full answer sheet. For an institution processing thousands of examination papers across multiple courses in a semester, these savings translate into faster student feedback and more time for faculty to focus on teaching.

The Limits AI Has Not Solved

The IIM Nagpur experiment is a pilot, not a deployment. Several well-documented limitations remain.

Subjective answers remain hard to grade reliably. AI systems trained on model answers can assess structured factual responses with reasonable accuracy, but management education involves long-form case analyses, strategic arguments, and ethical reasoning that resist standardized evaluation rubrics. A rubric can define what a "strong" answer looks like, but calibrating an AI to apply that rubric consistently across diverse student writing styles is an ongoing research problem.

Bias embedded in training data. AI models learn from historical graded answers. If those historical answers reflect evaluator biases — cultural, linguistic, or otherwise — the AI may replicate and scale those biases rather than eliminate them.

No regulatory framework exists. The UGC has not yet issued guidelines for AI-assisted grading in Indian universities. Without a regulatory framework, institutions using AI for grading operate in an ambiguous space that could create challenges for students seeking re-evaluation or legal redress for marks disputes.

Student awareness and consent. If an AI is evaluating examination work, students arguably have a right to know. The transparency standards for AI-in-the-loop assessment have not been established in the Indian context.

How This Differs from What Board Exams Are Doing

There is a distinction worth drawing clearly, especially given recent confusion in the media about CBSE's 2026 On-Screen Marking system.

CBSE's OSM for Class 12 boards involves digitally scanning answer sheets, transmitting them securely to evaluators' designated evaluation centres, and having teachers mark them on computers. The Tribune reported that CBSE explicitly clarified: "students need not panic — Class 12 board exams are not checked by AI." The intelligence in OSM is still human; the digital layer improves logistics, traceability, and error-checking.

What IIM Nagpur is attempting goes further: genuine AI reading and scoring. The distinction matters because the two systems carry different implications for accuracy, accountability, and student rights.

Why Universities Are Paying Attention

India has approximately 1,100 universities and 45,000 colleges. Each runs multiple examinations per semester. The total volume of answer scripts evaluated annually across the system runs into hundreds of millions of pages. The human bandwidth required — qualified teachers willing to spend days in evaluation centres — has become a genuine constraint.

CBSE had to issue directives to affiliated schools ordering them to release teachers for board evaluation duty, because schools were routinely not doing so. The evaluation workforce is finite, and it is competing with teaching duties, personal obligations, and in some cases reluctance to participate in centralized evaluation camps.

AI-assisted grading, if it matures to the point of reliability, addresses a structural problem that digital scanning alone does not. Scanning makes evaluation more secure and auditable. AI makes it faster and potentially less dependent on evaluator availability.

What Needs to Happen Before Wider Adoption

For AI grading to move beyond pilots at elite institutions to mainstream deployment in state universities and affiliated colleges, several things must happen:

The IIM Nagpur initiative is genuinely significant. It is one of the first serious institutional experiments with AI grading in Indian higher education, backed by the credibility of a premier institution. But it remains, for now, a proof-of-concept. The real test will come when the question is not "can AI grade papers faster?" but "does it grade them better, more consistently, and more fairly?"

Related Reading