A few days ago, I had an interesting interaction with some technology students. The discussion was primarily meant to be about the businesses these students can do. However, midway the discussion took an interesting turn and veered towards artificial intelligence (AI) – the business of AI; and AI in business. I would like to share a gist of the discussion with the readers.
In my view, there are primarily three kinds of businesses in the world.
1. Businesses designed to serve the needs of the people. Some of these businesses actually manufacture needs for people, mostly in the form of aspirations, and then serve to fulfil these needs/aspirations. Fashion, digital entertainment, luxury tourism etc. are some examples of such businesses.
2. Businesses to protect the lives and property of people. Some of these businesses, manufacturing of arms and ammunition, both threaten and protect the lives and property of people.
3. Businesses that serve the institutions and structures created to organize, govern, grow and manage people. Examples would be social & physical infra developers, universities, etc. There are certain businesses which are mostly based on systemic inefficiencies and corruption. Private coaching and general medical practice for example are two such businesses that should not exist in an efficient governance system.
These are of course broader and porous categories. For example, a hospital serves human needs and also anchors a healthcare system that is an institutional structure.
The framework of business
In my view, the business schools need to teach why businesses exist, not just how they make money. That is the more interesting question and the less taught one. Business schools spend a great deal of time on revenue models and competitive strategy and almost no time on the prior question: what is this business actually for?
The idea that a business can manufacture the very need it then profits from satisfying, is the defining feature of modern consumer capitalism. A clothing company does not simply clothe people. It constructs an anxiety — you are not dressed correctly, you are not current, you are not who you could be — and then sells the relief. Instagram does not simply connect people. It engineers a particular kind of loneliness that its own product then temporarily addresses.
Most of these businesses run on a variety of revenue and profitability models. Traditionally, the businesses that understood the human needs most and created a business model that would serve those needs in the best possible manner were the most successful. However, in the modern times, the businesses that inspire the human aspirations and manage to successfully fulfil these aspirations. Economists often ignore an uncomfortable principle - the demand curve is not always discovered. Sometimes it is built.
The business of arms is structurally about threat-and-protection simultaneously. It requires the world to be dangerous in order to justify its own existence.
Scarcity – a governance failure or arithmetic
In my view, in a country with good and equitably distributed public education, private tutoring would shrink dramatically.
A counter argument however could be that the problem in India is not merely that public schools are poor. The problem is that the gates are narrow. There are more people who want to be engineers and doctors and civil servants than there are seats in the institutions that confer those identities. Private coaching exists not just because teaching is bad — it exists because competition is intense and the stakes of that competition are high. You could fix every government school in the country tomorrow and still have a Kota. Scarcity is not a governance failure. It is arithmetic. And businesses that serve people navigating scarcity are not parasitic — they are, for better or worse, responding to something real.
It is a valid argument, but I would like to disagree.
What actually makes businesses succeed
Ideally, the businesses that understood the human needs most and created a business model that would serve those needs in the best possible manner were the most successful. Of course, it certainly does not mean that the most empathetic business always wins. The structural positioning of the business plays a major role in its success. The most durable competitive advantages in history have rarely been rooted in sensitivity to human desire. They have been rooted in control of chokepoints — physical, digital, regulatory, or social.
The uncomfortable truth is that many of the world's most profitable businesses have made their money not by understanding what people want, but by making it structurally difficult for people to want anything else.
Standard Oil did not succeed because John Rockefeller understood human needs more sensitively than his competitors. He succeeded because he controlled pipelines. The railways did not win because they grasped the human desire to travel — they won because once you owned the track, nobody could run a train without your permission. Google did not become the default portal to human knowledge because it was the most empathetic company — it became that because once you are the search engine everyone uses, you are the search engine everyone uses. Network effects are not about understanding needs. They are about structural position.
And so: where does AI go?
An interesting question was asked, “which of the three categories should we place the business of developing in selling Artificial Intelligence?"
The honest answer is that AI fits comfortably in all three of his categories — and that this is precisely what makes it unusual.
· It serves human needs directly: it reads your scans, drafts your contracts, translates your documents, helps you write the email you have been putting off for a week. Category one.
· It is already deeply embedded in the security and protection business: surveillance systems, autonomous weapons research, fraud detection, cyberdefence. Category two. And like the arms trade, it simultaneously creates and addresses certain threats — deepfakes being perhaps the most visible example.
· It is becoming infrastructure for institutions: government systems, healthcare bureaucracies, financial regulators, universities that now have to decide what a degree means when a machine can write the essay. Category three.
But here is what I think is actually interesting. AI is not just a business that fits across these categories. It is a technology that is changing what each category means.
When AI enters healthcare, the question of what general medical practice is for becomes more complicated. When AI enters education, the question of what a coaching centre is actually teaching becomes more complicated. When AI enters governance, the question of what institutions are even supposed to do becomes more complicated. It is not slotting into the existing structure. It is pulling at the joints.
The closest historical analogy I keep returning to is electrification. Electricity was not a business in any of the mentioned three business categories. It was a medium that transformed every business in every category — what manufacturing could do, what medicine could do, what governments could do, what entertainment could do. You could not place it in a taxonomy of 1880 because it was in the process of making that taxonomy obsolete.
I suspect AI is doing the same thing. Which means the question — which category does AI belong in? — may be a wrong question. The right question might be: what do the categories look like after AI has been inside them for twenty years?
I do not know the answer. But I think the students who ask this question are well ahead of most people who are simply using the product.