Do you remember your first interaction with ChatGPT? What about the last things you searched on Google or which Instagram Reels kept you watching or which advertisement caught your attention? You may not remember all of it. Your phone might. And that is where our story begins.
You order chai. Tap your phone against the QR code. Twenty rupees, gone, tea in hand – and by the time you take your first sip, you’ve already forgotten the transaction ever happened.
But somewhere, quietly, it hasn’t been forgotten at all. That four-second exchange just became a digital record: the amount, the time, the merchant and other transaction details. Multiply it by the roughly 22 billion UPI transactions India processes in a single month, and your chai stops being just chai. It becomes a line in a ledger you probably never think about.
It might be the world’s smallest invoice – paid in full, every single day, without you ever seeing the bill.
But what exactly is “data”? Is it spreadsheet rows or census tables or the marks on your report card that your parents definitely remember better than you do? But data is much simpler than that. Data is information. Your name is a form of data. Your photograph is a form of data. Your location is a form of data. Your examination marks are a form of data. A UPI transaction is data. Even the amount of time you spend watching a particular video can become data. One piece of information may tell us very little, but millions of pieces can reveal patterns. Sometimes, they can tell a story about you that you didn’t know you were writing – and definitely didn’t proofread.
The concept of data has existed thousands of years before modern technology, and in a broader sense, information itself has existed in nature in forms such as DNA. Over time, the ways in which humans have recorded and processed information have evolved—from cave paintings to paper, from tally marks to complex data structures, from Akbar’s zabt system to the modern-day census. Data has existed in every period in different forms, constantly evolving alongside us. Even the fundamental functioning of modern computing is based on one of the simplest ideas of representing data: binary, the system of 0s and 1s that lays the foundation for complex forms of computing.
How your feed starts to know you a little too well
Have you ever experienced that you watched a cricket video or searched something related to your favorite cosmetic brand and by evening, your entire feed is about cricket or about cosmetics.You never actually told Instagram or YouTube that you liked the sport or needed beauty products. You just watched or searched. Maybe you lingered a beat too long. Maybe you shared one. The platform noticed, because that’s the whole game: your behaviour becomes data, the data becomes a prediction, the prediction decides what you see next, and what you see nudges what you do next. Around and around it goes, less a feed, more a feedback loop with your name on it.
Artificial intelligence is what happens when you take that trick and put it on steroids. Feed a system enough data and it becomes remarkably good at spotting patterns in images, in language, in you. But here’s the catch nobody puts in the pitch deck: the system only knows what it’s been fed but it has zero instinct for whether what it’s been shown was fair to begin with.
In 2014, Amazon built an experimental AI hiring tool designed to score resumes one to five stars, like product ratings. It trained the model on ten years of resumes the company had received. Tech hiring skews heavily male, so most of those resumes came from men and the model concluded, correctly by its own logic, that male candidates were the safer bet. It would reportedly penalize resumes containing the word “women’s” or the names of certain all-women colleges, and it started favouring resumes that used masculine-coded verbs like “executed” and “captured.” Nobody programmed it to discriminate. It just did the math on the data it was given, and the data said: hire more of what we already have. Amazon eventually scrapped the project after losing confidence it could be made gender-neutral.
That’s the pattern, and it isn’t unique to hiring. India has run into its own versions of it: facial recognition tools used by police here have shown higher error rates on darker skin tones, and automated eligibility checks for welfare schemes have wrongly turned away people who genuinely qualified, because the data feeding the system was incomplete or skewed to begin with. The machine isn’t cruel. It’s just a very fast, very tireless copy of an old mistake – one capable of repeating itself millions of times without ever pausing to ask, “wait, should I be doing this?”
Data itself isn’t the villain of this story. The trouble starts the moment we forget that a dataset is a representation of reality, not reality itself and representations can leave people out, flatten the messy parts, or launder an old bias into something that looks objective simply because a machine said it.
Whoever holds the data, holds the room
So who actually owns all this? Mostly, a handful of tech companies and the arithmetic behind their size is almost quite simple. More users means more data. More data means more patterns and hence sharper predictions and these predictions are used to optimize user experience and ultimately better products. Better products pull in more users. Repeat. It’s less a business strategy and more a snowball rolling downhill, and the platforms already at the bottom of that hill have a head start nobody else can easily match.
The phrase ‘data is the new oil’ has been repeated so often that it has almost become a cliche. In fact data might be even more valuable than oil because oil may run out the moment you burn it. Data can be copied, recombined, reanalysed, resold, again and again, without ever wearing out. Its real value isn’t the raw fact itself (you bought this, you watched that). It’s what can be predicted from it: what you’ll buy next, where you’ll go next, what you’ll watch at 1 a.m. when you really should be asleep.
The biggest stakes in this story were never really about advertising, or convenience, or even money. They’re about attention. If you know what someone watches, what makes them angry, what keeps their thumb moving at midnight, you don’t just have information about them. You have a map of exactly their behaviour and as a result their brain. And a map like that is something you can use to influence them.
So the cycle is : predict what someone will click, and you can shape what they see. Shape what they see for long enough, and you shape what they think about. Do that across a whole population, and data has stopped being mere information. It’s become leverage – quiet, invisible, and remarkably effective.
And this is where data stops being merely commercial and starts becoming political. If you know what makes an audience angry, afraid, curious or hopeful, you have a map of its attention, and attention is something that can be directed. Governments, political campaigns and powerful organisations can use digital platforms to amplify certain narratives, target particular audiences and suppress or drown out others. China offers one of the best examples of how a state can control digital information and make it a tool of political power: by tightly controlling online speech and requiring platforms to remove or promote certain kinds of content, frequently critical of the regime. But the underlying concern isn’t uniquely Chinese. Wherever a government, company or organisation gains enough control over the information people see, the ability to influence an audience grows.
Which is really the whole case for calling data one of the defining currencies of our age. Not because it glitters like gold, but because whoever holds enough of it can move what millions of people spend their attention on, often without those millions ever noticing the transaction.
The real question is whether we’ll ever stop scrolling long enough to notice or just keep tapping “I agree.”