There’s no denying that Artificial Intelligence has become part of everyone’s life and made things easier. But it raises a real question: how is learning itself being affected as technology advances? We’re learning to build the technology, but not necessarily learning through the technology we’ve already built.

That question is quietly unsettling classrooms and faculty rooms across the country right now, and it isn’t about whether students are using AI. That much is already settled. According to the recent studies, more than 90% of learners have used AI in some form, and roughly nine out of ten students admit to using generative AI directly in their academic work. The real question universities are now grappling with is simpler, and far harder to answer: if a machine can write the essay, solve the code, and structure the report, how does anyone know whether the student behind that submission actually understands any of it?

To dig our what more is happening in the education sector, EduKida, collected a few quotes to vice-chancellors, deans, registrars and academic leaders across Indian higher education to understand how they’re rethinking assessment in an AI-saturated world, and while their institutions and disciplines differ, a common thread runs through nearly every conversation: the written submission, once the backbone of how universities measured learning, has quietly lost its power to prove anything at all.

The Problem With Asking for a Polished Answer
Dr. G. Pardha Saradhi Varma, Vice-Chancellor of KL Deemed to be University, frames the shift as a necessary evolution rather than a crisis. “The rapid growth of Artificial Intelligence is transforming how students learn, research, and complete academic assignments,” he says. “While AI tools can improve productivity and provide access to information, they also present a critical question for higher education: How can universities accurately assess whether a student truly understands a subject when AI can complete much of the work for them?”

His answer isn’t to reject AI outright, but to redesign what counts as proof of learning. “The future of evaluation lies in measuring not only what students can produce, but also what they can understand, explain, apply, and demonstrate independently,” he explains. That means leaning more heavily on live problem-solving sessions, viva voce examinations, practical demonstrations, and case-based discussions, formats where a student has to think on their feet rather than submit something polished from home. “A student who understands a concept should be able to explain its principles, defend their approach, respond to unexpected questions, and apply their knowledge to a new situation, even without relying entirely on an AI-generated response,” he says. For universities, he adds, the real challenge “is therefore not simply to detect AI use, but to design assessments where genuine understanding becomes visible.”

“An Open-Book Environment on Steroids”
Dr. Radhika Srivastava, President & CEO of the Fortune Institute of International Business (FIIB), pushes this idea further, arguing that trying to catch AI use is the wrong fight altogether. “When AI can draft a solid essay in seconds, traditional submission-based assessments lose their diagnostic power,” she says. “The solution isn’t banning AI or relying on unreliable detection software; it is evolving how we evaluate mastery.”

At FIIB, she describes AI as “an open-book environment on steroids,” a phrase that reframes the entire problem. “If an assessment can be completed entirely by a prompt, it was measuring task completion, not understanding,” she says. Her institution has responded with what she calls process-oriented evaluation, three techniques in particular: bringing back viva voce and real-time defence, where students explain and defend their reasoning live; introducing applied friction, deliberately ambiguous case studies with incomplete data that force judgment rather than pattern-matching; and process tracking and auditing, where students submit their original AI prompts and document how they critiqued and refined the output. “Ultimately, we aren’t testing whether students can generate answers anymore,” she says. “We are testing whether they can evaluate, critique, and take responsibility for those answers.”

Testing for Judgment, Not Just Knowledge
That idea, that the goal now is judgment rather than recall, comes up again and again. Mr. Chirag Gupta, Vice-President of IMS Noida, describes it as a shift from static knowledge to cognitive agility. “If an AI can complete an assignment end-to-end, we are evaluating the software’s output rather than the student’s actual understanding,” he says. “Generative tools have effectively automated standard responses, but what they cannot replicate is human judgment when conditions change unexpectedly.”

Across IMS Noida’s IT, business management and journalism programmes, faculty have started introducing live curveballs mid-assessment, a sudden data drop, an altered client brief, a piece of code that breaks in front of an evaluation panel. “A student cannot rely on a pre-generated AI prompt when they must pivot their strategy on the spot, justify their choices during a viva, or fix a live error in front of an evaluation panel,” Gupta explains. “When answers are instantly accessible, real understanding reveals itself in how a student handles friction and defends their logic under pressure.”

Dr. Shailesh Rastogi, Director of Badruka School of Management (BSM), sees the same principle applying to management education specifically. “If AI can seamlessly complete an assessment, that test was merely measuring lower-order recall, not genuine executive capability,” he says. His institution has built its response around three pillars: live Socratic evaluation and interactive vivas that force students to field counterarguments in real time; unstructured business cases “filled with live market noise and human ambiguity,” designed so that pattern-matching alone fails; and prompt auditing, where students submit their full AI interaction logs alongside a reflective critique, identifying exactly where the model “hallucinated, lacked context, or failed strategic depth.” As Rastogi puts it, the goal of modern management education is no longer to train students to synthesise data into standard reports, “it is to cultivate leaders who can interrogate, refine, and strategically apply knowledge after AI has already done the heavy lifting.”

From Grading Answers to Grading Reasoning
Dr. Tanya Singh, Dean Academics at Noida International University, ties this back to the numbers driving the whole conversation. With the vast majority of students already using generative AI in their coursework, she argues academic leadership has no real choice but to rethink what assessment is meant to measure in the first place. “Conventional tasks focused only on the final written piece may no longer reflect students’ abilities,” she says. Her recommendation echoes what’s emerging as a near-universal playbook: viva-voce examinations, active problem-solving, practical demonstrations, and application-based evaluation. “A student may prepare a solution with the help of AI,” she notes, “but the ability to say why the solution works, to challenge the assumption behind the reasoning, or to apply it in a new situation offers greater evidence of real understanding.” Her broader point is one of balance rather than resistance: “The idea should not necessarily be considered something that schools should get rid of altogether. Instead, educational institutions should teach students how to use it properly.”

What a Faculty Member Is Actually Listening For
Few describe the practical mechanics of this shift as clearly as Dr. Gurbinder Singh, Registrar of Thapar Institute of Engineering & Technology. “AI has changed what a take-home assignment tells us,” he says. “A polished report or a clean piece of code is no longer proof that a student understands the subject, because generative AI can now produce both in minutes. The solution is not to ban the tools. It is to change what we ask the students to work on.”

At Thapar, that’s translated into live problem-solving, vivas built around the simple but revealing question “why did you choose this approach?”, and application-based tasks that ask students to apply a concept somewhere new rather than repeat what they’ve read. In programming courses specifically, a student might submit code for an assignment, only to have a faculty member change the requirement on the spot and ask them to modify it live, explaining every decision as they go. “The signal we look for is reasoning,” Singh says. “A student who understands the subject can explain a trade-off, spot the flaw in a wrong solution and adapt when the conditions change. A student who has only submitted generated work usually cannot do this under questioning.” He’s careful to frame the intent behind all this correctly: “For universities, the shift is not about catching students out. It is about making sure that a degree still reflects what a graduate actually does.”

Certifying Thinking, Not Output
Viraj Sagar Das, Pro-Chancellor of BBD University, brings the conversation to what might be its clearest single statement of purpose. “If a student can generate a well-structured report or assignment with a few prompts, the report or assignment itself stops being a reliable signal of understanding,” he says. “What we can’t outsource to AI is a student’s ability to explain their reasoning, defend a decision under questioning, or apply a concept to a problem they haven’t seen before. This is exactly where assessment now needs to move.”

Das is careful to draw a distinction that runs through nearly every perspective in this piece: using AI well and understanding a subject deeply are not the same skill, and universities need assessments that can actually tell the two apart. “This isn’t about distrusting AI or treating it as a threat,” he says, noting that students absolutely should learn to use these tools well, since that’s the reality of the workplace they’re entering. But faculty, he adds, need to get comfortable asking “harder and more improvisational questions in real time, rather than relying solely on submitted work.” His closing thought captures where this entire conversation seems to be heading: “The University’s job is not to certify whether a student can produce an output but to certify that they can think for themselves. In the AI-saturated world that we are currently living in, interactive assessment may very well be the only reliable way left to do exactly that.”

A Quiet Redesign, Already Underway
What emerges from these conversations isn’t panic, and it isn’t resistance to AI either. It’s a fairly clear-eyed recognition, shared across engineering institutes, business schools and universities alike, that the tools measuring learning haven’t kept pace with the tools now doing the learning’s heavy lifting. The fix isn’t detection software or outright bans. It’s bringing back something that predates AI entirely: sitting a student down and asking them to explain, defend, and apply what they claim to know, live, unscripted, and without a prompt to lean on.

Share on Facebook

Post on X

Follow us

Save