A lever, a microwave and a machine that answers back: what makes AI different

Give me a long enough lever and a place to stand, Archimedes is supposed to have said, and I will move the world. It is one of the oldest boasts about technology, and it captures what most of our machines have always done: they make us stronger, faster or more precise, but they do not decide anything. The lever does not choose what to lift.

Artificial intelligence is the first technology for which that sentence is no longer completely true. That is what I want to think through in this post: what actually separates AI from the tools that came before it, why our scepticism about it is healthy, and why, in a country like India, the conversation has to start somewhere very different from where it starts in a country like Canada.

What a tool is

Think about a lever, a microwave, a hammer, a knife. Each one takes something a human wants to do and amplifies it or rather simplifies it. The lever multiplies force. The microwave heats food faster than a fire and is more resource effective too. A hammer drives a nail that a fist could not.

Three things are true of all of them. First, their behavior is fixed by their design: a microwave does exactly what its engineers built it to do, every time, and an engineer can explain why. Second, the purpose comes entirely from the person holding it. The same knife cuts vegetables or hurts someone; the difference between a tool and a weapon lies in the hand and the intention, not in the steel. Third, they work on the physical world: matter, heat, force, distance and so on.

Even complicated machines, a car, a printing press, a calculator, follow these rules. A calculator does arithmetic faster than you or I can, but it does not decide which sum is worth doing.

What makes AI different

Modern AI breaks each of those three rules a little.

  • Its behavior is learned, not designed line by line. Nobody wrote down the rules a language model follows to answer a question. It learned patterns from enormous amounts of data (data on which these models are trained), and even the people who built it cannot fully say why it produced a particular answer.
  • It makes judgements. A lever never decides anything. An AI system decides which word comes next (contextuality is in-built), which loan application looks risky, which X-ray looks abnormal, which video you see after this one (lord help us, with our Instagram algorithms). Part of the purpose has moved from the hand into the machine.
  • It works on the mind, not on matter. AI writes, summarises, translates, explains and persuades. It operates in the same space where we think, learn and form opinions, which means it can shape the person using it in a way a hammer never could.
  • It is general-purpose and spreads fast. A microwave only heats food. The same AI model can draft a legal letter, tutor a child in algebra and write computer code, and it can reach hundreds of millions of people in months rather than decades.

That is why the old tool-or-weapon question gets harder. With a knife, we know whom to hold responsible. With AI, responsibility is spread across the people who collected the data, the company that trained the model, the institution that deployed it and the person who asked the question. When something goes wrong, it is genuinely difficult to say whose hand was on the handle.

Every earlier machine extended what our bodies could do. AI is the first that starts to do some of our thinking, and that is a different kind of change.

We have always been sceptical, and we should be

None of this means we should panic. Humans have greeted almost every new technology with suspicion. In Plato’s Phaedrus, Socrates worries that writing will weaken memory and give people the appearance of wisdom without the substance. Scholars feared the printing press would flood the world with too many books. Weavers in nineteenth-century England, the Luddites, smashed the machines they believed would destroy their livelihoods. When pocket calculators arrived, teachers argued that children would never learn arithmetic. Many of us remember being told the internet would rot our brains.

Some of these fears were exaggerated, and some were partly right: the Luddites did lose their trades, and writing did change how we remember. But in the end people embraced these technologies, because they made life easier, safer and more livable. Books spread knowledge beyond monasteries. Machines lifted much of humanity out of back-breaking labour. Mobile phones put banking in the pockets of people who had never seen a bank.

The lesson I take from this history is not “the sceptics were wrong”. It is that scepticism is useful when it asks the right questions: Who benefits? Who is left behind? Who is accountable when it fails? What are we giving up? Those are exactly the questions we need to be asking about AI, and asking them is not the same as rejecting it.

Why AI asks us to rethink society

Because AI works on cognition, it touches the institutions that are built around human thinking. Schools and universities assume that the work a student hands in shows what the student can do. Workplaces assume that writing, analysis and coding are scarce skills. Courts and newspapers assume that a document, a photograph or a voice recording is evidence of something real. All of these assumptions are now under pressure at the same time.

Rethinking society in the light of AI means asking, deliberately, how we want to teach and assess, which work we want humans to keep doing, how we will know what is true, and who gets a say in how these systems are built. If we do not ask, the answers will be decided for us by whoever builds and sells the technology.

Two very different starting points: Canada and India

I live and work in Canada now, and I grew up in India. Watching the AI conversation in both places has made one thing very clear to me: it is not the same conversation.

In a country like Canada, where most people have reliable electricity, clean water, healthcare and a decent internet connection, the debate is about productivity, jobs, ethics and regulation. Those are real questions. But they are the questions of people whose basic needs are already met.

In India, many people are still struggling to make ends meet and do not have reliable access to basic resources: nutritious food, clean water, healthcare, a good school nearby. For a family deciding between school fees and medicine, a discussion about large language models is not just distant; it can feel almost insulting.

At the same time, India has made a serious push into the digital world. The government launched the Digital India programme in July 2015 with the aim of turning the country into a digitally empowered society and knowledge economy, and digital literacy schemes like PMGDISHA have targeted rural households. In 2024 the Cabinet approved the IndiaAI Mission, with a budget of about ₹10,372 crore over five years, focused largely on computing infrastructure, startups and building Indian AI models.

Connectivity has grown remarkably: India had roughly 958 million active internet users in 2025, with rural India now the majority and growing fastest. But the same report shows how uneven that access is. Nearly one in five internet users goes online through someone else’s phone, and most of them live in villages. In a country of about 1.4 billion people, several hundred million are still not online at all.

So I think the AI conversation in India can begin in the light of Digital India, but it will be, and should be, delayed compared to richer countries, because it has to start one step earlier. Before we argue about what AI should do, we have to make sure people know what it is.

First, AI literacy

AI literacy does not mean everyone learning to build neural networks or their own LLM’s. It means knowing what these systems are, what they are good and bad at, when to trust them, when to check them, and how to protect yourself from being misled by them. Someone who receives a convincing voice note from a “relative” asking for money urgently needs AI literacy far more than they need an AI startup.

Some places I think we should start:

  • Teach in people’s own languages. Most AI tools work best in English, but in India that is only the beginning of the problem. The country has 22 scheduled languages and hundreds more that people actually speak, so this needs real thought. Some would argue that everyone should simply learn English or Hindi. But we learn best in our mother tongue, especially those of us who are not specialists. If we want AI literacy to reach ordinary people, we have to remove the language barrier. This connects to the old debate about teaching in the mother tongue, but that is a topic for another day.
  • Start with teachers. No programme reaches children if their teachers are confused or afraid. Train and trust teachers first.
  • Design for shared and basic phones. If a family shares one phone, lessons and tools have to work on that phone, with patchy data, not on a laptop with fast Wi-Fi.
  • Use the public access points that already exist. Schools, libraries, panchayat offices and Common Service Centres can become places where people try AI tools safely, with someone to explain them.
  • Put women and girls at the centre. Women are often the last in a household to get a device and the first to be targeted by online abuse, so any literacy effort that ignores them will widen the gap. In many Indian families, it is mothers who spend the most time at home with young children and who teach them first, from early childhood. If we want a digitally literate generation, we have to reach and teach the mothers first. As the old saying goes, educate a man and you educate one person; educate a woman and you educate a whole family.
  • Make good tools free or cheap. If useful AI sits only behind expensive subscriptions, it will deepen inequality rather than reduce it. We have to make AI, and digital goods in general, far more accessible. In India this gap is especially stark. In the metros, almost everyone is online and digitally literate. But much of the country still lives with deep inequality in access to basic resources, and that has to be addressed first. An AI literacy programme designed for Bengaluru or Mumbai will not work in a village without a reliable internet connection.

Where I want to go from here

A lever moves rocks. AI is beginning to move ideas, decisions and opportunities. Whether it makes life more livable for most people, as the best technologies of the past did, depends less on the technology itself than on who understands it and who gets access to it. In the next posts I want to dig into AI literacy in classrooms, how assessment has to change, and what all of this means for a country where a single exam can involve millions of students.

I would love to hear from you, especially if you teach, study or work in India: what does AI look like where you are?

Sources

Comments

Leave a Reply

Is this your new site? Log in to activate admin features and dismiss this message
Log In