AI has changed the question, schools must change the answer

This article is authored by G Sridhar, founder, Deeksha Group of Institutions.

Learning, as we have known it for over two centuries, has been centred around knowing things. Gathering information was the fulcrum on which academic outcomes turned.

AI (Photo credit: Unsplash)

There was a reason for this: information richness was career insurance. Creativity and originality were always valuable, but information was the low-hanging fruit with a long shelf life.

Things changed when the internet reached our pockets; and now, with generative AI, popularised as LLMs, the information-intensive education system has been disrupted forever. The pace of this change is unprecedented. According to the Stanford AI Index Report 2025, AI systems are now matching or surpassing human performance on an increasing number of benchmark tasks, while adoption across education and workplaces has accelerated dramatically. The question is no longer whether AI will become part of learning—it already has.

Today, writing a coherent essay, getting a solution to the toughest differential equation, or drafting a debate opening is not a cognitive challenge at any level. And this is not to be seen as a crisis of copying or cheating, but as the new normal of processing knowledge.

Now, when “What do you think”, not "What do you know?", is the real differentiator, we have to begin rethinking the structure of schooling itself, from textbooks to homework to exams.

Does this mean that information is meaningless? Not really. Foundational learning and a critical mass of information are needed to be creative; however, beyond a threshold, information has drastically diminishing payoffs.

A student who learns biological classification by rote will know how the animals are grouped. Here are two possibilities: In the process of learning, he might become awed by the diversity of the living world and then become interested in evolutionary biology, or he may attempt a project in conservation. Alternatively, a crowded syllabus overwhelms him, he learns by rote and never gets curious. Far too many students take the second route, and marks incentivise it.

The result is an entire generation that wonders what sine and cosine even mean and why they ever bothered with the gravitational constant or medieval history. This generation is also prone to believing that academics work in silos and since subject streams do not meet in exams, they also do not intersect in real life. This is also where education is beginning to diverge from employment. The World Economic Forum's Future of Jobs Report 2025 identifies analytical thinking, curiosity, resilience, AI literacy and lifelong learning among the fastest-growing skills employers will seek over the next five years. Yet most classrooms continue to reward recall over inquiry.

With AI on the scene and demonetisation of the information currency, we must urgently begin to embed curiosity in our curricular design. Create classrooms where asking questions and judging answers is the primary activity. Learning spaces where information is used exclusively to spark self-directed research and analysis.

When curiosity is the curriculum, silos vanish. Subject delineation will still have a functional utility, but students won’t build their academic identity around them. It will be a kind of academic cosmopolitanism, where everything is interconnected. Students naturally find connections between ecology and physics and mathematics and music, and all along they will be assisted, not challenged by AI.

What naturally follows when academic objectives are revised in this way is an emphasis on experiential and problem-solving, leading to students proposing solutions to bad AQI or understanding Bangalore’s lake restoration work.

Assessments are the pivot on which the brave new education turns. An AI-driven economy is reshaping the professional world by demanding and rewarding the ability to originate ideas, judge them, and collaborate with AI to build on them.

We will still test information, but not through direct recall. It will be done by demanding application, depth of understanding, and the ability to work through a real, unresolved problem, not recite a solved one.

For instance, asking students to define photosynthesis can be replaced with why it cannot be mimicked at scale.

Similarly, an essay on the climate is pointless beyond grade 4; presenting the climate crisis as a Shakespearean tragedy is a perfect cross-disciplinary ask.

Marks will still matter, but they will stop being the be-all and end-all. Students who thrive will pair sharp thinking with adaptability, sound judgement, and the ability to work well with AI and collaborate with peers.

Academics, ambition, and legacy of a generation are shaped by the questions it faces. The problems today are of a globally disruptive scope. A changing climate challenging the growth model of economies, technology redefining ethics and relationships, growing inequality, all topped up with AI disruption. These have no ready answers, and to prepare children to find meaning in this world needs them to engage with open questions, not revel in resolving the solved ones.

To do that, we have to first inculcate judgement, which is the capacity to find information, weigh it, decide its reliability and value, and build something new from it.

We have argued that the first step is curiosity, but let us qualify that now by saying that it is half the story. Asking a sharp question is critical, but it is only the beginning; knowing whether to trust the answer is the culmination; no syllabus has caught up to it yet. AI's response to an original question can be wrong, biased, or badly argued, and a brilliant question answered badly and accepted without scrutiny is counterproductive. Judgement is the ability to argue and compel an LLM to refine the response; it goes far beyond saying “something else” or “I don’t like it.”

Picture a student who asks why we gained Independence in 1947, and not in any other year. Now, this is a thoughtful question. The AI's answer comes back fast, citing Britain’s exhaustion after World War II, ongoing Indian resistance, and sentiment against colonialism. It is a confident answer, but it is still really an argument open to challenge. Only a student willing to argue back can pull an original thesis from it rather than settle for the first draft. That willingness to challenge a confident answer is judgement, and it is hard to teach precisely because it cannot be memorised.

Schools around the world are still catching up. UNESCO has noted that fewer than one-third of countries have developed formal guidance on the use of AI in education, leaving teachers and institutions to navigate profound changes without clear policy frameworks. That makes judgement—not prohibition—the more sustainable response.

Schools are already grappling with poor AI answers submitted as assignments. Who is responsible when a lazy AI answer finds its way to the evaluator? We can’t pretend that we have a definitive answer because the question never came up. How do we resolve this? By encouraging judgement and AI response curation. AI restriction has its place, but it can’t be the key policy.

AI has made it easy to arrive at an answer without earning the thinking behind it. What schools owe their students now is a deliberate decision to teach children to ask sharper questions and distrust easy ones. That is the new answer to the question that AI has already changed.

(The views expressed are personal)

This article is authored by G Sridhar, founder, Deeksha Group of Institutions.