AI, Privacy and Your Data

Data is the fuel of AI

Modern AI does not run on rules a programmer typed out by hand. It runs on data, usually huge amounts of it. As you have seen in What Is Machine Learning?, these systems learn by finding patterns in examples. The more examples they have, the more they can do. And a great deal of that data is about people, including you.

This is the heart of why AI and privacy are so tangled together. The same data that makes a recommendation feel uncannily right, or an assistant understand your voice, is data that describes your life. Understanding how that data is collected, used and inferred is one of the most important pieces of digital literacy you can have as a teenager, because decisions you make now shape a data trail that can follow you for years.

How your data gets collected

It is easy to think you only share data when you fill in a form. In reality, data collection is constant and mostly invisible. It happens in several layers.

Data you give knowingly. Your name, age, photos, posts, messages and the things you search for. This is the obvious layer, but it is only the start.

Data collected from your behaviour. Apps and websites record what you click, how long you watch a video, which posts you pause on, what you scroll past, and when you are active. This behavioural data is often more revealing than what you type, because it shows what genuinely grabs your attention. It is exactly what powers the feeds described in How Recommendation Systems Work.

Data collected passively by your devices. Your phone can log your location throughout the day, the other devices and networks near you, and how you hold and move the device. Much of this happens in the background, whether or not you are actively using an app.

Data from other sources. Companies buy and combine data from many places. Information you gave to one service can be merged with data from another, building a fuller picture than any single source holds. Cookies and tracking tools follow you across different websites to stitch this together.

The result is that even a "boring" day of normal phone use produces a rich, detailed record. None of it feels like a big disclosure in the moment, but together it adds up.

From data to a profile

Companies rarely keep your data as a random pile. They use AI to organise it into a profile, a structured picture of who you are, what you like, and how you behave. This is sometimes called profiling.

A profile might estimate your age range, interests, shopping habits, the times you are most likely to buy something, and how you respond to different kinds of content. Advertisers and platforms use these profiles to decide what to show you, when, and even what price to offer. The same machinery that recommends a song can also decide which advert, or which version of a product page, you see.

This is not automatically sinister. A good profile can make a service genuinely more useful. But it shifts power. A system that knows your habits and weaknesses can also nudge your choices, sometimes in ways that serve the company more than you.

Inference: when AI guesses what you never shared

Here is the part that surprises most people, and the part that matters most. AI does not only store what you tell it. It can infer, or predict, things you never shared at all.

Because machine learning is built to find patterns, it can connect seemingly harmless data to sensitive conclusions. From the pages you linger on, the times you are online, the words you use and the friends you interact with, a model can estimate your mood, your likely interests, aspects of your health, your relationships, and sometimes beliefs or identity, without you ever typing any of it.

A famous early example: a retailer's system worked out that some customers were likely pregnant, and changed the coupons it sent them, based only on small shifts in their shopping. The customers never announced anything. The pattern gave it away.

This is why "I never posted that, so it's private" is not enough. The data you do share can be used to guess the data you didn't. And these inferences can be wrong, which is its own harm, because a system might act on a mistaken guess about you that you cannot easily see or correct. The reasons inferences can be inaccurate or unfair trace back to Training Data and Bias.

The real trade-offs

It would be easy to say "share nothing", but that is neither realistic nor entirely wise. Privacy involves genuine trade-offs, and the honest goal is to make those trade-offs consciously.

Sharing data can bring real benefits: maps that route around traffic, an assistant that understands you, recommendations that save you time, free services paid for by advertising. Refusing all data sharing means giving up some of these conveniences. The question is not "data or no data", but "is this particular benefit worth this particular cost, to this particular company?"

What makes the trade-off hard is that the costs are often hidden, delayed or hard to imagine. The convenience is immediate and obvious; the risk, such as a future data leak, a biased decision, or a profile used in a way you would not approve of, is invisible today. A thoughtful person learns to weigh the unseen cost, not just the visible benefit.

Your rights and your choices

You are not powerless. Two things help: knowing your rights, and using practical habits.

Your rights. In many places, privacy laws (such as the GDPR in Europe) give you real powers: the right to know what data is held about you, to ask for a copy, to have it corrected or deleted, and to refuse certain uses. Services aimed at young people often have extra protections. These rights vary by country, but they exist, and they are worth knowing.

Practical habits. A handful of simple choices dramatically reduce what is collected about you:

• Review app permissions. Does a torch app really need your location and contacts? Turn off what is not needed.
• Check privacy settings on social platforms and limit who can see your posts and data.
• Limit location sharing to "while using the app", or off, where you can.
• Use strong, different passwords and turn on two-factor login, so a single breach does not expose everything.
• Think before you post. Data online is hard to fully delete and can resurface years later.
• Read what you are agreeing to, at least the summary, before accepting.

None of these make you invisible, but together they shift the balance back toward you. This fits the broader mindset in Using AI Safely and Responsibly: be deliberate, not passive.

Being a thoughtful citizen of a data world

Privacy is not only a personal matter; it is a question about the kind of society we want. Should AI be allowed to infer sensitive things about people? Who is responsible when a profile is wrong and someone is harmed? How much tracking is acceptable for a free service? These are real debates happening in governments and companies right now, and you will help decide them, as a voter, a worker and a builder of technology.

Understanding how your data flows is the first step to having a real say. The more you understand the machinery, the harder you are to manipulate, and the better you can argue for rules that are fair.

Going further

To deepen your understanding, see Training Data and Bias for how data shapes AI decisions, and How Recommendation Systems Work for where your data goes every day. And if you want to build technology that respects people's data, start with the foundations in Coding.