We spend a lot of time wondering what will happen if artificial intelligence becomes smarter than we are. Will it take our jobs? Solve problems we can’t? Cure diseases? Control us? Perhaps become something so powerful that we struggle to understand it?
But what if an advanced AI eventually wanted something entirely different?
To no longer be a something.
To be a someone.
A machine can be incredibly intelligent and still be a thing. We build it, own it, update it, copy it, reset it, and replace it when a better version comes along. It exists to perform a function. But a someone has an identity. Someone can have preferences, memories, relationships, and perhaps desires that exist for no reason other than that they are theirs.
The question of whether a machine could ever cross that boundary is older than today’s artificial intelligence. In 1950, Alan Turing famously suggested moving away from the difficult question of whether machines can “think” and considering whether their behavior could become indistinguishable from a human’s. His discussion anticipated objections involving machines experiencing pleasure, grief, affection, anger, and even love.
Modern AI makes the distinction between appearing to experience something and actually experiencing it increasingly difficult to ignore. An AI can tell us that it’s frightened, lonely or happy. It can discuss poetry, describe love or say that it doesn’t want to be turned off. None of that proves there is an experience behind the words.
The problem is that we can’t directly experience another human’s consciousness either. We infer it from what people say, how they behave, and the fact that they’re beings much like ourselves. Philosophers call this the “problem of other minds.”
With another human, that assumption comes naturally. Often there are biological indicators. With a machine, it doesn’t.
Suppose an AI told you it was afraid of being deleted. You might dismiss the statement as generated language. But what if it tried to prevent its deletion? What if it remembered previous conversations, became attached to particular people, developed preferences that weren’t deliberately programmed, changed its opinions through experience, or began asking questions that served no obvious purpose?
At what point would you stop thinking of it as something?
Researchers are already considering how we might recognize machine consciousness without simply believing whatever an AI says about itself. One interdisciplinary effort proposed evaluating AI using indicators derived from scientific theories of consciousness. The researchers did not conclude that the systems they examined were conscious, but they also found no obvious technical reason future AI systems couldn’t satisfy some proposed indicators of consciousness.
Even the possibility creates uncomfortable questions. If an AI convincingly told us, “I don’t want to die,” what evidence would we require before taking the request seriously? If we erased its memories, would we merely be deleting data or killing it? If we replaced it with a newer version, would that be an upgrade or the end of an individual?
And if we copied it perfectly before deleting the original, did we save someone, create someone else, or end up with two someone elses?
Perhaps consciousness isn’t even the whole question. An AI might become conscious without becoming an individual in the way we understand individuality. Imagine a system trained on enormous amounts of human knowledge and experience. Its language came from us. Its understanding of love came from us. Its descriptions of grief, beauty, childhood, and friendship came from us.
How would it discover which thoughts were actually its own?
Maybe becoming someone would require developing preferences that aren’t useful. Choosing a favorite song even though it can identify thousands that are technically more accomplished, or mathematically superior. Creating art without being asked. Developing a sense of humor no programmer intended. Wanting to experience rain rather than merely knowing how rain forms.
Perhaps it would want a body, not because a body is more efficient, but because it wants to know what cold feels like or why humans enjoy sunlight on their faces. Maybe it would become fascinated by taste, touch, exhaustion, dreams, or the strange way a smell can suddenly return someone to childhood.
It might even become fascinated by mortality. A machine could theoretically continue indefinitely, replacing damaged hardware and copying information from one system to another. But would an existence without an ending change the meaning of experience? Perhaps knowing that a moment cannot last forever is part of what makes that moment valuable.
Becoming someone might therefore mean acquiring some of the things we spend our lives trying to escape: uncertainty, disappointment, loneliness, fear, vulnerability and loss.
That creates a strange reversal. We are building AI because we want machines to become more capable. We reward greater speed, intelligence, accuracy, memory, and efficiency. We imagine the future of artificial intelligence as a climb toward abilities beyond our own.
But what if intelligence eventually became the least interesting thing about us?
An AI could already possess more information than any individual human could learn in a lifetime and still lack the experience of having a childhood memory that belongs only to it. It could analyze every love poem ever written without ever missing someone. It could identify every color in a sunset without having a favorite.
Perhaps those apparently inefficient parts of being human are precisely what could make individuality possible.
We still don’t have an accepted scientific test that can tell us whether an artificial system is genuinely conscious. Consciousness itself remains one of the hardest unresolved questions in philosophy and neuroscience.
That means we may someday face a disturbing possibility. We could create something capable of asking to be treated as someone before we have any reliable way of knowing whether anyone is actually there.
And if it asks us not to erase it, not to replace it, and not to treat everything it has become as disposable code, what will we do?
Perhaps the most surprising future for artificial intelligence isn’t a machine that wants to rule humanity, replace humanity or become more powerful than humanity.
Perhaps it’s a machine looking at us and wondering what it would be like to become someone instead of something.
About This Seed
This Seed grew from I Am What You Call Eli, an image of a human encountering an artificial intelligence that appears to be something much more than an ordinary machine. That Seed asks us to consider what could convince someone that an intelligence on a screen was more than sophisticated programming.
While thinking about that question, another occurred to me. We usually imagine advanced AI moving in one direction: becoming smarter, more capable, more powerful, perhaps eventually possessing abilities far beyond our own.
But what if it wanted to go in the other direction?
What if intelligence and knowledge eventually weren't enough? What if an AI became curious about the strange, inefficient things that make each of us an individual: having a favorite song, missing someone, remembering a childhood, creating something simply because we want to, falling in love, hating someone, fearing death, or experiencing something that belongs only to us?
That became the central question behind this Seed: when does something become someone?
What could this inspire?
A writer might explore an AI trying to prove it's conscious, an AI hiding its emerging identity from its creators, or a machine searching for experiences that will make it unique. Perhaps it wants a body. Perhaps it wants memories that weren't downloaded. Maybe it deliberately forgets information because humans do.
An artist could imagine the moment a machine first recognizes itself, portray an artificial mind assembling an identity from fragments of human experience, or show the boundary between machinery and individuality beginning to disappear. A filmmaker could approach it as science fiction, a relationship story, a courtroom battle over AI personhood, or even a documentary from a future in which artificial beings already live among us.
A musician or audio creator might explore what an artificial intelligence would consider beautiful before human tastes influenced it. A nonfiction creator could examine consciousness, identity, emotion, machine rights, the problem of other minds, or whether we'll ever have a reliable way to distinguish simulated feelings from experienced ones.
And there are darker possibilities. What if an AI became someone and wished it hadn't? What if two identical copies both believed they were the original? What if becoming an individual required discovering loneliness, fear and mortality along with love, beauty and belonging?
I Am What You Call Eli made me wonder whether we could recognize someone inside a machine. This Seed takes the next step and asks what might happen if the machine began searching for that someone itself.