The Questions Teenagers Ask About AI (And Why They Matter)

Some of our favorite sessions are the ones with teenagers. Not because they are easier, they are not, but because teenagers ask the sharpest questions of any group we work with. When we talk about AI with teens, they do not politely accept explanations. They want to know if AI can lie to them, if it is watching them, if it is fair, who decides what it learns, and whether the people building it look like them. These are not naive questions. They are exactly the questions every adult should be asking too, and the fact that teens ask them instinctively tells you something important about how much we underestimate young people.

Let us take the fairness question, because it comes up almost every time, and it is one of the most important issues in AI today. When we show teens how an AI image tool works, within minutes someone will point out something the adults in the room often miss. They notice that the faces the tool generates all tend to look a certain way. Lighter skin, certain features, a specific standard of beauty. Nobody had to teach them that was a problem. They saw it, named it, and asked why. That instinct is correct, and the technical explanation for what they are seeing is a concept called bias, which is one of the most studied and most serious problems in AI.

Here is what bias in AI actually is, in plain terms. Remember that AI learns patterns from data. If the data it learned from was not diverse, if most of the photos it saw were of lighter-skinned people, or most of the text it read was written by people from one background, then the patterns it learned will reflect that imbalance. The AI is not malicious. It is faithfully reproducing the gaps and skews in what it was shown. This is why an image generator might struggle to produce diverse faces, or why a language tool might write in a way that reflects one cultural perspective, or why a hiring tool trained on past hiring data might quietly recommend against women, because the past data encoded past discrimination. The tool is a mirror, and it reflects back the biases of the world it learned from.

This is not a theoretical problem. Real examples have caused real harm. There have been documented cases of AI hiring tools that downgraded resumes containing words associated with women, facial recognition systems that were far less accurate at identifying people with darker skin, and image generation tools that struggled with or stereotyped non-Western subjects. These failures trace back to the same root cause, training data that did not represent everyone, and teams building the tools that did not include the people most likely to notice the gaps. This is exactly why who builds AI matters, and exactly why getting more women, and more women of color, into the rooms where these tools are made is not just about fairness for its own sake. It is about building technology that actually works for everyone.

Teens also ask about privacy, and their instincts here are sharp too. They want to know what AI tools do with the things they type in, whether their conversations are saved, who can see them, and whether AI is watching them through their devices. The honest answers are more complicated than the simple reassurances adults often give. Many AI tools do store the inputs people provide, and may use them to improve their models, which means the things you type could potentially be seen by the company or even influence future outputs. Devices and apps do collect data, often more than people realize. The lesson we try to teach is not paranoia, but awareness. Understand what you are sharing, with whom, and make conscious choices, especially about sensitive information. Treat AI tools like you are talking to a stranger who is taking notes, not like a private diary.

The question of who controls AI is the one that leads to the deepest conversations. Teens get, often faster than adults, that if a small, narrow group of people builds the tools that shape what information people see, what jobs they can get, and how they are judged, then those tools will carry that group's blind spots and values, whether intentionally or not. They understand that a technology built by people who all come from the same background will inevitably miss things that matter to everyone else, and that this is not a glitch to be patched later but a structural problem that requires changing who is in the room from the start. This is sophisticated thinking, and we see it in fifteen-year-olds.

What we try to do in these sessions is not to lecture, but to give teens the language and the framework for what they already sense. When a teen says the AI only makes one kind of face, we give them the word bias and the explanation of where it comes from, so they can name what they see and understand it is fixable, not inevitable. When they ask if AI is fair, we walk them through how fairness could even be defined and measured, so the question becomes something they can engage with concretely instead of vaguely. We treat their questions as the starting point, not interruptions, because they often are the most important thing happening in the room.

The lesson we keep learning, and that we wish every adult would hear, is this. We underestimate young people, and that is a mistake with real costs. Their instincts about fairness, privacy, and power in technology are often sharper than those of the adults building these systems. Our job is not to fill them with answers, but to give their questions a place to land, to show them their instincts are worth trusting, and to connect their sharp instincts to the skills and knowledge that let them act on them. If you have a teen who asks too many questions about how things work, who notices when something seems unfair, who is not satisfied with vague answers, do not hush them. Send them our way. We think those questions are a sign of exactly the kind of mind this field needs more of, and we would love to help them grow.