June 20, 2025 — In an era defined by unprecedented technological leaps, one reality looms increasingly large: while artificial intelligence (AI) reshapes industry, governance, entertainment, and everyday life, a vast majority of society remains deeply illiterate about what AI truly is — and isn’t.
Professor Bent Flyvbjerg’s recent article, AI as Artificial Ignorance, offers a timely reality check that tempers both the hype and the paranoia surrounding AI’s perceived omniscience. He bluntly reminds us that the large language models (LLMs) powering tools like ChatGPT are, at their core, sophisticated text generators — not reasoning engines, not truth-tellers, but statistical parrots repeating patterns extracted from vast troves of text.
Adding to this cautionary chorus, retired Brigadier Amit Kathpalia, an engineer and management educator, recounted a telling incident: when he asked ChatGPT a question about contract law, it confidently served up an incorrect answer. When confronted with court judgments contradicting its claim, the AI scrambled to self-correct, awkwardly spinning excuses like a cornered human debater. Kathpalia has since issued an unvarnished warning: “A fool with a tool is still a fool… and AI can make that fool more dangerous.”
The Call for AI Literacy
These anecdotes speak to a larger truth now echoed by technologists and educators alike: if India’s leap into the AI era is to be transformative and equitable, citizens must understand not just how to use AI tools, but how to think critically about them.
A recent editorial in a leading Indian daily pointed out that, just as conventional literacy rose from 12% in 1948 to over 75% today — fuelling India’s economic and social mobility — the 21st century demands a new kind of fluency: AI literacy.
But what does this really mean in practice?
What AI Really Is — and Isn’t
At its heart, artificial intelligence is not a monolithic mind but a suite of mathematical methods that allow computers to learn patterns from data. In simple machine learning, an algorithm ingests known examples, identifies trends, and predicts outcomes for new, unseen data. This has revolutionised tasks like weather forecasting, clinical diagnostics, and targeted advertising.
Yet not all AI is created equal. Classic computing performs precise calculations step by step. The human brain, by contrast, excels at recognising images, reading emotions, or catching nuance — tasks where logic alone falters. To bridge this gap, engineers developed neural networks — digital architectures that mimic the way biological neurons process information.
For instance, to train a computer to distinguish cats from dogs, programmers feed it thousands of labelled images. The neural network gradually tweaks millions of internal weights to minimise mistakes — a process called back-propagation. After enough examples, the system becomes surprisingly good at spotting a Labrador or a Persian, even in new photos it has never seen before.
Large language models push this idea further: instead of labelling cats and dogs, they predict which words are statistically likely to follow a given prompt. Trained on mountains of books, articles, and websites, they can churn out plausible paragraphs, write poems in Shakespeare’s style, or draft legal memos — but they don’t understand any of it.
This is the kernel of Flyvbjerg and Kathpalia’s warning: the machine doesn’t “know” truth; it knows patterns. And when these patterns draw from flawed or biased data, or when conflicting facts exist, the AI may confidently pick the wrong path — with no intuition to check itself.
The Double-Edged Sword of AI’s Power
Despite its limitations, AI’s practical uses are transformative. Early in the last century, human “efficiency experts” painstakingly optimised factory production using pen, paper, and primitive math. Today, AI does this on a scale and speed impossible for humans. From optimising airline routes and dynamic pricing to real-time fraud detection and tailoring ads for every smartphone user, AI’s commercial benefits are undeniable — and profitable.
In fields like healthcare, autonomous driving, and disaster prediction, AI has already crossed thresholds once thought unreachable. At the same time, it has enabled deepfakes, manipulative profiling, and sophisticated fraud — giving it a darker reputation in public discourse.
Yet, as history shows, society usually finds ways to adapt to the perils of powerful technologies. The printing press spread both enlightenment and propaganda; the internet brought knowledge to billions while creating fertile ground for misinformation. AI’s trajectory may be no different: the challenge is to educate users to discern hype from reality, benefits from threats.
The Case for Teaching AI Literacy
So what would true AI literacy look like for India? A recent policy report, Mapping AI in India, argues that computational thinking — problem-solving, abstraction, and designing logical workflows — must become core to the school curriculum. This does not mean everyone must become a programmer. Rather, it means understanding how AI tools work, where they can fail, and how to wield them wisely.
While digital natives — the smartphone-savvy 12-year-old — may have an intuitive sense of what AI can do, many adults remain in the dark, especially where education systems are overstretched. In some states, basic math and science competence among school teachers remains worryingly low; one recent assessment showed government high school teachers averaging just 22% and 26% on science and math pre-tests, respectively.
Without raising these foundations, talk of building an AI-literate population rings hollow.
A Cautious but Confident Future
Ultimately, the goal is not to fear AI, nor to worship it as an oracle, but to treat it as what it is: a powerful but limited tool. If society can build critical awareness, strengthen schools, and cultivate a culture that questions instead of blindly trusting the machine, then India can reap AI’s benefits while avoiding its worst pitfalls.
Flyvbjerg’s blunt reminder rings true: “A fool with a tool is still a fool.” But with AI literacy, tomorrow’s citizens may become wise enough to ensure that the tool empowers, not misleads.