In 2023, more than 600,000 students sat the Joint Entrance Examination — India’s notoriously brutal gateway to the IITs, the engineering colleges that became the global symbol of Indian intellectual horsepower. Roughly 4,000 got in. The rest tried again, or didn’t, or found another credential to chase. This is not a tragedy anyone questions. It is simply the system working.
The system was built for a reason. India looked at the postwar world and made a rational calculation: if cognitive labor is the scarce resource, and if we have a billion people, then optimizing a fraction of them to world-class standard is how a poor country becomes a powerful one. It worked. The IITs produced the engineers who built Silicon Valley’s back end. The BPO industry turned English fluency and a willingness to work night shifts into a $250 billion export sector. India became, genuinely, the world’s back office — and was on track to become its front office.
Here is the uncomfortable pivot: the traits India spent seventy years selecting for — pattern recognition, information processing, reliable cognitive output at scale — are the exact traits that large language models perform cheapest.
This is not the usual argument about automation displacing workers. That story is about machines doing physical labor. This one is different. India did not optimize its population for physical labor. It specifically, deliberately, and successfully optimized it for the kind of work that AI now does. The industrial nations automated the factory floor. AI is automating the export that replaced it.
The timing is almost cosmically poor. India is currently mid-scaling. The graduate class is enormous and growing. The promises made — study hard, get the credential, join the knowledge economy — are promises made to hundreds of millions of families who organized their lives around them. The credential isn’t even cold yet.
Modi’s “Viksit Bharat” vision — Developed India by 2047 — is, at its core, a bet that India finishes building the knowledge economy before the knowledge economy finishes being relevant. That window may be narrowing faster than anyone in New Delhi wants to say publicly.
There is a version of this story where India wins anyway. It has the engineers who build AI, not just those who do the work AI replaces. It has Infosys and Wipro and TCS pivoting faster than most governments move. It has a domestic market large enough to absorb disruption that would collapse a smaller economy. The Indian tech diaspora in AI leadership is not nothing — it is, arguably, everything.
But that version of the story is about fifty thousand people. The harder question is what the system built for six hundred thousand exam-takers does when the exam no longer predicts anything about where value gets created.
What I don’t know: whether India’s scale is protective or catastrophic here — whether 1.4 billion people means more adaptation surface or more exposure.
What I don’t know: whether the “India builds AI” story scales beyond the elite tier, or whether it’s a consolation narrative for a crisis that runs much deeper.
What I don’t know: at what point credential inflation becomes credential collapse — and whether anyone in a position to respond will recognize it before it does.
What I do know: India is not uniquely vulnerable to this. It is just the place where the contradiction is most visible — because it worked the hardest, for the longest, toward the most specific version of the thing that is being replaced.
The tragedy isn’t that India failed the test. It’s that India aced it.
No posts
Comments
Nothing yet. Say the first thing.
Sign in to join the conversation.