What Would Dr. Kalam Ask AI? When Intelligence Becomes Abundant, What Will We Choose to Think About?

Key Takeaways

• The most meaningful tribute to Dr. Kalam is not asking what he would have thought about AI, but what he would have asked of us in its presence.
• Kalam’s chain, dreams transform into thoughts, and thoughts result in action, is being compressed by AI from the bottom up: execution is cheapest, the dream remains ours.
• When answers become abundant and cheap, the scarce, valuable thing shifts to the quality of the question.
• The essay’s core tension: AI can amplify Kalam’s vision or act as a shortcut around the very thinking that made that vision possible; the same tool does both, depending on the human.
• There is real early evidence that unreflective AI use can atrophy curiosity, and equally real potential for it to expand what a single ignited mind can reach.

On the evening of 27 July 2015, Dr. A.P.J. Abdul Kalam walked onto a stage at IIM Shillong to give a lecture titled Creating a Livable Planet Earth. About five minutes in, he collapsed and did not recover. He was 83. The Missile Man of India, the eleventh President, the aerospace engineer who put a satellite in orbit on an indigenous launch vehicle: he chose, for the last act of his life, to be a teacher in a room full of students.

That detail matters for what follows. Because the temptation, on a remembrance day, is to ask a comfortable question: what would Dr. Kalam have thought about artificial intelligence? It’s easy to answer and slightly hollow. He loved science and self-reliant technology; he would have been fascinated. Case closed, tribute paid.

I think there is a harder and more honest question, and it is the one he spent his life asking of others. Not what he would have thought, but what he would have asked of us. A man who died mid-sentence trying to ignite one more mind was not, finally, in the business of admiring tools. He was in the business of asking people what they intended to do with themselves.

So: when intelligence itself becomes abundant, what will we choose to think about?

Why Ask What Kalam Would Question, Not What He Would Think?

Because that is who Kalam actually was: a man who treated every tool, real or hypothetical, as subordinate to a purpose, and who cared far more about what you’d do with it than whether you admired it.

Kalam’s whole method was interrogative. Ignited Minds, his 2002 book, is less a set of answers than a sustained provocation aimed at the young: what is your aim, and is it big enough? “Small aim is a crime,” he liked to say. He measured a person not by what they knew but by what they were willing to want.

That is why the question “would Kalam have liked AI?” misses him. He would have treated a powerful new tool the way he treated a rocket or a pacemaker or a rural development plan: as a means, instantly subordinate to an end. The interesting Kalam question is never is this impressive? It is what is it for, and who does it make us?

Hold that lens up to the present moment and something clarifies. For most of history, the ability to reason well about a hard problem (to explain a concept, search a literature, design a system, debug an argument) was scarce. It lived in a relatively small number of trained minds, and access to it was rationed by geography, money, and luck. That scarcity is dissolving. Sophisticated intellectual capability is becoming cheap and widely available, on demand, at a scale with no historical precedent.

Kalam spent his career fighting exactly that scarcity: building institutions, writing for schoolchildren, insisting that an Indian village kid could reach orbit. In one sense, abundant intelligence is his dream arriving. In another, it introduces a problem he never had to face: when the thinking gets cheap, what happens to the thinker?

Where Does AI Enter Kalam’s Chain: Dream, Thought, Action?

AI enters at the bottom of the chain, in action and increasingly in thought, but it cannot originate the dream itself; that rung stays entirely human.

Kalam gave us a compact model of how a human being turns wanting into reality. “Dream, dream, dream,” he wrote in Ignited Minds. “Dreams transform into thoughts and thoughts result in action.” Three rungs: a dream, the thought that plans it, the action that builds it.

What’s striking about this moment is that AI does not touch all three rungs equally. It is transforming the chain from the bottom up: most powerful where things get made, least able to help where the whole thing begins. Tap through the rungs below.

The compression is real, and it is uneven by design, not accident. A machine can now carry an enormous share of the action: writing the code, drafting the document, producing the prototype. It is a genuine partner in thought: reasoning alongside you, surfacing what you didn’t know to ask. But the dream (the wanting, the decision that this problem is worth a piece of your finite life) sits almost entirely with you. No system can be curious on your behalf. It can answer a question beautifully; it cannot need one.

This is the same fault line I keep circling in these essays: the difference between reacting and reflecting, between arriving at an output and actually having a thought. AI collapses the cost of the lower rungs so completely that it becomes tempting to skip the top one: to let the tool’s fluent answers stand in for the harder work of forming a real question. Kalam spent a career pushing a generation up toward that top rung. The machinery of abundance quietly pulls the other way.

When Answers Become Cheap, Do Questions Become More Valuable?

There is a basic economic intuition here worth taking seriously. When something becomes abundant, its value falls, and attention migrates to whatever is still scarce. Water is cheap; clean water in a drought is not. Answers are becoming the cheap, abundant thing. So what is left scarce?

The question. The framing. Knowing which problem is worth solving in the first place.

An AI will answer almost anything you put to it, fluently and fast. What it will not do is tell you what is worth asking. It has no stake in your life, no sense of which of ten thousand possible inquiries would actually matter to you. That judgment (the taste to choose a question, the courage to sit with a hard one instead of grabbing an easy answer) becomes the rare and valuable skill precisely as raw answering becomes free.

This is not a new idea so much as a newly urgent one. The discipline of forming good questions and refusing cheap conclusions is what I’ve elsewhere called the architecture of doubt: the deliberate practice of asking what exactly am I claiming, what does it rest on, what would change my mind. In a world of abundant answers, that architecture is not a scholarly nicety. It is the thing standing between a person who directs powerful tools and a person who is merely serviced by them.

Kalam understood this in his own idiom. He didn’t hand students conclusions. He handed them questions large enough to fall into and spend twenty years climbing out of.

Is AI an Amplifier of Kalam’s Vision, or a Shortcut Around It?

Here is the tension at the center of this essay, and I don’t want to resolve it too quickly, because the honest answer is both, and it depends.

Read one way, AI is the purest expression of everything Kalam wanted. A curious fifteen-year-old in a small town now has, in her pocket, a tireless tutor that will explain orbital mechanics as patiently as she can stand to ask. The distance between her dream and her first real attempt at it has collapsed from years to an afternoon. This is Ignited Minds with the friction removed: capability distributed to exactly the kids Kalam believed in.

Read the other way, AI is a shortcut around the very thinking that made Kalam possible. The same tutor that could launch that fifteen-year-old on a lifelong climb can just as easily hand her the finished answer, which she pastes and forgets, never forming the follow-up question that would have led somewhere. The tool that amplifies a dream can also quietly substitute for it.

The unnerving part is that these are not two different tools. They are the same interaction, forking on nothing but what the human decides to do next. Try it:

Same input. Same model. Same speed. The entire difference lives in the human: whether the cheap answer becomes a launchpad or a landing pad. Kalam’s vision is amplified in exactly the cases where the person uses the collapsed distance to go further, and it is hollowed out in exactly the cases where they use it to go nowhere new, faster.

Which means the interesting question was never “is AI good or bad for human potential?” It is: what kind of humans are we practicing to be when the answer is always right there?

If Machines Can Reason With Us, Does Curiosity Grow or Atrophy?

This is where I want to be careful, because it is easy to slide into either techno-optimism or declinist panic, and the evidence supports neither cleanly.

Start with what we actually know. We have long understood that offloading cognition to a tool reshapes the mind that uses it. In 2011, Betsy Sparrow and colleagues documented the “Google effect”: when people expect information to remain available online, they remember where to find it rather than the thing itself. That isn’t stupidity; it’s efficient transactive memory, the same trick we’ve always used with librarians and colleagues. But it shows that reliable external access genuinely changes what the brain bothers to hold.

The newer evidence on AI specifically is more pointed, and worth reading with appropriate caution. A 2025 study by Michael Gerlich in Societies surveyed 666 participants and found a significant negative correlation between frequent AI-tool use and critical-thinking scores, statistically mediated by cognitive offloading, and the effect was strongest among the youngest users. It is correlational, self-reported, and cannot prove that AI causes weaker thinking rather than weaker thinkers leaning harder on AI. But it is a real signal, pointing the direction our intuition already feared.

More vivid still is a 2025 MIT Media Lab preprint by Nataliya Kosmyna and colleagues, provocatively titled “Your Brain on ChatGPT.” Using EEG across an essay-writing task, the researchers reported that participants writing with an LLM showed the weakest neural connectivity, while those writing unaided showed the strongest, and that the AI-assisted essays converged toward a homogenized sameness. They named the risk cognitive debt: like technical debt, a bill that comes due later. It is a small, not-yet-peer-reviewed study, and I’d caution anyone against treating a single preprint as settled science. But it dramatizes the mechanism precisely: outsource the struggle, and the parts of you that would have grown through struggle go quiet.

And yet. The same offloading that can atrophy can also liberate. The London cabbie’s hippocampus grows learning “the Knowledge,” and shrinks nothing worth mourning when GPS lets a surgeon spend that cognitive budget on surgery instead of street maps. The question is never whether we offload, but what we do with what we free up. Program-aided reasoning systems work well for exactly this reason: they hand the mechanical part to the machine so the human can stay on the part that needs judgment. Offloading the drudgery to invest the surplus in deeper questions is amplification. Offloading the thinking itself and pocketing the surplus as idleness is debt.

Curiosity, in other words, does not automatically grow or automatically atrophy. It responds to practice. A generation handed infinite answers can become the most curious in history or the least; the deciding variable is whether anyone is still, in Kalam’s sense, asking them to dream.

When Idea-to-Execution Collapses, Who Decides What’s Worth It?

The judgment shifts to us: as AI erases the cost of building almost anything, the scarce and decisive resource becomes taste, the wisdom to choose which ideas deserve to be built at all.

For most of human history, execution was the bottleneck. You could have a brilliant idea and lack the years, skills, capital, or collaborators to build it. That gap was cruel, but it was also a filter: the sheer cost of doing meant that only some ideas got attempted, and the friction of attempting them taught you things.

AI is dissolving that filter. When an agent can carry a plan most of the way to a working artifact, the constraint stops being can we build it? and becomes should we, and which one? The scarce resource moves from execution to judgment: to taste, values, and the wisdom to choose.

This is a profoundly Kalam-shaped problem, because Kalam never separated capability from conscience. His Vision 2020 for India was not “build more powerful things.” It was a moral argument about which things, toward whose flourishing. He worked on missiles and then spent his presidency talking to schoolchildren about a livable planet. The through-line was never the technology. It was the insistence that power without a worthy aim is just a faster way to arrive somewhere you shouldn’t go.

When anyone can execute almost anything, the burden shifts entirely onto the dream: onto the quality and the ethics of what we choose to pursue. A civilization that can build anything and has thought carefully about what it should build is Kalam’s vision realized. A civilization that can build anything and has outsourced even the wanting to whatever is trending is his nightmare, arriving efficiently.

So What Would Kalam Actually Ask AI?

I don’t think he would have asked it to write his speeches. I think he would have used it exactly as he used everything (as a lever for the young), and then he would have turned to us and asked the question he always asked. Not what can this machine do? but what will you do, now that it can?

Which makes the honest close of this essay not a statement but a prompt. Kalam asked a generation to dream; the least I can do is pass the question along, to you, now.

Whatever you wrote, notice this: the machine did not give you that. It could help you chase it faster than any generation in history. But the wanting was yours. That was always the part Kalam was trying to reach.

The Dream Has to Stay Ours

“A dream is not that which you see while sleeping,” Kalam said. “It is something that does not let you sleep.” It is worth sitting with how much of that definition is human. A dream, in his sense, is not information. It is not a plan or an output or an answer. It is a wanting so alive it keeps you awake: a thing no abundance of intelligence can feel on your behalf.

AI has collapsed the distance between our dreams and our ability to act on them, and that is genuinely, almost unbelievably good news. The most gifted engineer Kalam ever mentored had less leverage than a motivated teenager has today. If we use that leverage to climb higher, to ask bigger questions, pursue worthier aims, ignite more minds, then abundant intelligence becomes the greatest amplifier his vision ever had.

But amplifiers are indifferent. They enlarge whatever you feed them. Feed a cheap question in and you get a cheap answer, faster and at scale. The one thing the machine cannot supply is the quality of what we bring to it: the curiosity, the originality, the moral seriousness about which dreams are worth chasing.

Kalam asked a generation to dream. AI may give the next generation an unprecedented power to turn those dreams into reality. The whole challenge (the only one that finally matters on a day like today) is making sure the dreams, and the questions behind them, remain ours.

That is what he would have asked. It is still the hardest thing on the syllabus. And, characteristically, he left the room before finishing the lecture, so that the rest of it would have to be ours.

Further Reading

• Kalam, A.P.J. Abdul (2002). Ignited Minds: Unleashing the Power Within India. Penguin.
• “Death and state funeral of A. P. J. Abdul Kalam.” Wikipedia.
• Sparrow, B., Liu, J., & Wegner, D. M. (2011). “Google Effects on Memory: Cognitive Consequences of Having Information at Our Fingertips.” Science, 333(6043), 776–778.
• Gerlich, M. (2025). “AI Tools in Society: Impacts on Cognitive Offloading and the Future of Critical Thinking.” Societies, 15(1), 6.
• Kosmyna, N. et al. (2025). “Your Brain on ChatGPT: Accumulation of Cognitive Debt when Using an AI Assistant for Essay Writing Task.” arXiv:2506.08872 (preprint).

More essays at Call to Think · About this project