How to go beyond AI in education

The Hindu

24-05-2025

AI prompting is powerful. It accelerates work, scales ideas, and connects information. But without context, curation, and personal agency, it leaves you painfully mediocre. To go beyond AI, focus on what it cannot do and work on the unpromptables in learning.
You can prompt AI to write an essay, summarise a book, solve Maths, or code. It delivers fast and polished results. But being fast is not mastery and being polished need not be original. The Prompt Paradox is the illusion that all knowledge can be generated through proper prompting. Prompts are useful but insufficient when dealing with context, curation, and metacognition. As you see, learning also calls for things that cannot be prompted.
Real-world experience
Reading about climate change may provide information, ideas, and statistics, but understanding its impact on a fishing village involves knowledge of rising sea levels in a lived context. Facts remain hollow without the weight of real-world experience. Not all contexts can be lived. In the same way, not all knowledge can be prompted.
Prompts can describe experiences, but they cannot replicate the visceral nature of embodied learning, such as dancing, enjoying the taste of a mango, or struggling to learn cycling. Similarly, prompts can provide ethical frameworks, but they cannot truly simulate the lived experience of moral deliberation.
Spotify changed how we listen to music. Instead of following artists or albums, many of us get algorithmic playlists based on mood or other preferences. This is convenient, but it flattens diversity and misses the context. In the same way, in AI-driven learning, when we receive personalised content, the context is collapsed, and intellectual diversity is reduced. Processing knowledge in fragments strips it of its depth.
Active curation
Real learning, like real music, demands active curation. Algorithms can assist but should not decide. Otherwise, we risk being trapped in a loop of the familiar — never challenged, never changed. Repeated exposure to second-hand information builds an illusion of knowledge, not expertise. Curation is knowing what to leave out to focus on what is important. The more we rely on AI-automated promoting, the less agency we apply in choosing what to engage with as learners.
A student researching war can gather hundreds of articles in seconds and summarise its dimensions, but forming a perspective requires human discernment, which cannot be prompted. Much online course content is often recycled, leading to diminishing returns as each iteration merely repeats its predecessors. Prompting cannot solve the problem of digital regurgitation, which clutters the Internet with zero originality. Both learners and educators should be aware that true personal voice emerges from reality; not from algorithms or rehashed data. If avatars blur reality, students risk losing personal agency, becoming trapped in algorithmically defined personas.
Struggling with mathematical proof builds mental resilience. However, when AI provides quick fixes, it short-circuits this essential struggle, holding back the natural cognitive growth. While AI can mimic thought patterns using graph-of-thought prompts, it lacks self-awareness. Metacognition — thinking about your own thinking — is key to evaluating and planning how you learn, in such situations. But here is the thing: you cannot just prompt it. To go beyond AI, focus on what it cannot do.
Before querying AI, journal your thoughts to clarify understanding and gaps in understanding. Engage with the real world. Read, discuss, and experience to go deeper. Actively filter and interpret information to develop discernment, rather than passively consuming online content.
Use AI as a starting point, not a conclusion. Integrate ideas instead of substituting them. Combine prompting with your insights to contribute meaningfully rather than echoing. While AI delivers answers, only you can assess their significance; true understanding arises from your interpretation, not mere data shuffling where endless rephrasing saps originality.
Challenge AI by crafting questions that expose its limitations and biases. Run Bias Audits by generating AI content on controversial topics and scrutinising inherent prejudices. Let the learner understand the fragility of the prompted 'knowledge'.
Practise cognitive disobedience: question algorithmic suggestions and uphold your human agency. Unquestioning reliance on algorithmic authority will soon be a serious AI issue in learning. Paradoxically, these practices will make you a better and more responsible AI prompter.
You can prompt for answers. You cannot prompt for understanding. True learning happens where AI stops and you move from prompting to understanding with learner autonomy. There, you rely on context, metacognition, personal expression, and the constructive struggle of meaning-making. Surrounded by algorithms, if we fail to confront potential biases and ignore the unpromptables in learning, we reduce ourselves to an average learner. That is the act of critical engagement with AI. It separates thinkers from mere users of AI.
Views expressed are personal
The writer is Deputy Secretary, University Grants Commission.