BHUBANESWAR: For most of the history of global student recruitment and mobility, the largest decision the students and their family made was shaped by whoever happened to be accessible to them. An uncle, neighbour, or friend of a friend who had once flown abroad and come back with opinions. Knowledge was local, and so, very often, was destiny.
That world is ending quietly, on millions of screens after midnight.
Today an eighteen-year-old in Indore can open a chatbot and ask what once required a human consultant and a little bit of waiting: Where should I study? What should I study? And what will it do for my life? And the answer arrives in seconds, in fluent English, with an unprecedented confidence.
India sent roughly 1.2 million students abroad in 2025, according to the Ministry of External Affairs, a number that fell by close to six percent as visa regimes tightened across the West. Yet the way students decide changed completely.
India now accounts for the largest single share of ChatGPT mobile users on earth, more than one in eight globally. And according to a recent Leap study of over three million students, roughly three in four Indian students now use AI to choose their programme. The first counsellor an Indian student meets is no longer a person. It is a model.
AI collapses the old bottleneck, where a student compared five to seven universities because that was all one office could realistically explain. Now a student weighs course, cost, scholarships, and post-study work rights across a dozen countries before breakfast, asking harder and more financial questions than any brochure ever encouraged. It is why interest in destinations like Germany surged in a single year, while the familiar Anglophone giants lost their monopoly on attention.
But the deeper shift is not what students can see. It is who the machine chooses to cite.
When a student asks a model for “the best” or “the safest” university, the answer draws disproportionately from the primary source: the institution itself, its course catalogues, its research output, its official record.
Superlatives act as filters that resolve toward institutions the model can classify under those words. For two decades, servicing companies and listing portals were the discovery layer, the toll booth every student passed through. Models do not need that toll booth. They return the institution directly. For the first time, institutions hold a structural advantage in discoverability, and the intermediaries who built businesses on being found first must now earn their place another way.
This is reshaping the human layer too. The student who walks into a counsellor’s office today has often already read more, compared more, and shortlisted more than the counsellor expected. That student challenges the advice, checks it against the machine, and occasionally knows the visa rule better than the person across the desk. The comfortable counsellor who traded in information alone is being exposed. The sector is being pushed, hard, to raise its game: to offer judgment rather than lookup and nuance rather than brochures.
This is the work my team does at Crizac. We are investing in AI on the other side of that equation: helping institutions and partners improve their own discoverability, automating the marketing content that decides whether a model cites them at all, and building certifications that lift counsellors from information-givers into genuine advisers. The aim is not to compete with the machine. It is to make sure the humans in the chain are still worth consulting once the machine has spoken.
Machine is not always right!
Because the machine is not always right, and it never sounds unsure when it is wrong. Visa rules and admission rules change without announcement to the models and are specific to each country, exactly the kind of fact a model trained months ago delivers with false confidence.
A student who trusts a hallucinated deadline or a course that no longer runs discovers the error only when it is expensive.
And there is a limit no upgrade will fix. For a very large number of Indian families, a degree abroad is not a purchase but a collective act.
Parents remortgage, relatives contribute, a decade of savings rides on one ticket. A machine can compare two universities. It cannot carry the weight of that consequence.
So, the honest answer is both. AI has made the average student far better informed and slightly more exposed. It has handed institutions a new advantage in being found, and handed counsellors a new challenge to deliver high-quality service as it is no longer limited to sharing the facts.
Facts and reassurance
AI can hand a student a thousand facts in a second. It cannot offer reassurance. It can describe how a university looks on paper. It cannot tell a student how it will feel to sit in that lecture hall, far from home, and slowly come to belong there. Information is what the machine gives. Meaning, story, and reassurance are what a human counsellor can give.
So, the question is no longer whether students will use these tools. That is settled. The question is who around them rises to meet a better-informed generation.
Crizac, the only listed company in this space, is choosing to lead that path, building proprietary models that put stronger technology in the hands of the people who guide students, so that discovery stays fast but the experience stays whole.
(The author is CIO of Crizac Limited)