What Makes Us Human in the Age of AI?

I build software, and I work with artificial intelligence. I have watched a system turn a sentence into an image, summarize a long document, or produce a plausible answer in seconds. The first reaction is often delight. The second is usually a more difficult question: if a machine can do this, what exactly was the human doing?

It is tempting to answer quickly. We might say that machines only follow instructions, or that they will eventually do everything we can do. Both statements sound decisive. Neither is much help. The technologies are changing, and so are the definitions we use to describe them. I find it more honest to stay with the uncertainty for a moment.

Intelligence is not one thing

For a long time, intelligence was treated as if it could be measured by a small set of familiar abilities: remembering facts, solving puzzles, calculating quickly, using language. AI systems now perform impressively across some of those tasks. They can identify patterns in data, generate fluent prose, and assist with work that once required hours of human effort.

That should make us reconsider what we mean by intelligence, not pretend the accomplishments are trivial. But performance on a task does not settle every question about the experience behind it. A system can produce a useful explanation of grief without having lost someone. It can describe the ocean without feeling cold water or waiting for a wave to arrive. The distinction between generating a description and inhabiting an experience is meaningful, even if we do not yet have a complete account of consciousness.

I do not claim to know whether an AI system could be conscious. We do not have a simple test that settles what subjective experience is, even in other humans. Fluency can persuade us that someone understands, but words alone cannot answer every question about what is happening inside a mind. The uncertainty calls for humility in both directions: we should avoid inventing inner lives for systems, and avoid declaring that difficult questions have already been solved.

The human parts of making

People sometimes worry that using AI makes creative work less human. I understand the concern, but I think the more useful question is where judgment enters. A tool can generate options; someone still has to decide what is worth making, what feels true, what is appropriate for the people who will use it, and what should be left out.

When I build software, the code is only part of the work. There is the person trying to complete a task while tired. There is a team trying to maintain a system after its original authors have moved on. There are tradeoffs between speed, reliability, privacy, and cost. A model can help with implementation or analysis, but the responsibility for those tradeoffs belongs to the people who choose how to use it.

Creativity has never been a solitary act of producing something from nothing. It is also selection, revision, attention, and the willingness to follow an idea beyond its first convenient form. AI can participate in parts of that process. It may make certain kinds of expression more accessible, help people explore unfamiliar approaches, or give a useful starting point when the blank page feels difficult. It can also flood a space with material that looks finished before anyone has asked whether it matters.

That is why efficiency cannot be our only measure of progress. A faster answer is valuable when it gives people time for better work. It is less valuable when speed becomes the reason to remove care, context, or human contact. The point of a tool is not simply to make a task disappear. It is to help us decide what deserves our attention next.

Connection is more than output

Some of the most important things we do are not impressive because of their output. Listening to a friend does not produce a document. Sitting with someone in uncertainty may not solve anything. Remembering how a person takes their tea is not a productivity achievement. These acts matter because of the relationship in which they happen.

Technology can support connection. It can help someone communicate across distance, translate between languages, or find information that would otherwise be inaccessible. It can also simulate the shape of attention without sharing the vulnerability and responsibility that make attention meaningful between people. The interface may feel personal; the relationship still depends on what it is, what it can do, and what the person using it believes it to be.

Human meaning is not a leaderboard. It comes from the lives we experience, the relationships we sustain, the obligations we accept, and the worlds we choose to shape together. A tool can expand what is possible without deciding what is worthwhile.

Choosing what to build

As someone who builds software and explores AI, I feel both curiosity and responsibility. Curiosity asks what a system can do. Responsibility asks who benefits, who bears the risk, and whether the system is being used in a way people can understand and question. Those questions do not slow progress for its own sake. They help distinguish a useful capability from a careless deployment.

The next decade will not be shaped only by what machines can do. It will also be shaped by what companies reward, what institutions permit, what communities refuse, and what each of us chooses to make ordinary. We can use technology to remove tedious barriers, widen access to knowledge, and help people do work they value. We can also use it to demand more output, hide accountability, and treat human attention as an inefficiency to be optimized away.

The same capability can serve very different purposes. That is why the human question is not only what machines might become. It is what we want our tools to make of us. Do they help us become more attentive, more capable, more generous with our time? Or do they train us to accept whatever is fastest as if it were automatically better?

I cannot give a final definition of what makes us human, and I am not sure a final definition would serve us well. The question may be more useful when it stays open, reminding us that intelligence is not the whole of a life and that invention carries choices inside it.

Perhaps our task is not to prove that no machine could ever resemble us. It is to build tools that leave room for the parts of life that cannot be reduced to output: care, responsibility, wonder, and the freedom to choose what matters. The future will be shaped by what we can build, but also by what we decide should remain human.