What AI taught me about the soul — On intelligence, mortality, and why only beings who can die are capable of meaning
Published on Medium: 2026-05-05 — https://medium.com/@danithaca/what-ai-taught-me-about-the-soul-on-intelligence-mortality-and-why-only-beings-who-can-die-are-43c2f4ed336f
I’ve been spending a lot of time working with AI lately — using it to write, to think, to solve problems I used to spend hours on alone. It’s useful in ways that still catch me off guard. But somewhere in that process, a strange question started nagging at me: what exactly is intelligence? And once I started pulling that thread, a deeper one came with it — what is the difference between an intelligent agent and a soul?
I don’t have a definitive answer. But I’ve been thinking about it long enough that I want to try putting it into words.
What AI can do: the essence of intelligence
Large language models are, at their core, extraordinarily capable problem-solvers. Tell an AI what you need — analyze this dataset, translate this contract, write this code — and it will draw on vast learned knowledge plus trained reasoning to find a path to the answer. That capacity, applying accumulated knowledge and logical inference to reach a goal, seems like a reasonable definition of intelligence itself. Intelligence is not magic. It is the ability to navigate from problem to solution cleverly.
By that definition, AI is already intelligent. And it may soon be more intelligent than any individual human across most domains that can be specified as goals.
What AI cannot do: know why it should care
Here is where things get interesting. Ask an AI to help solve global poverty and it will produce thoughtful, well-sourced analysis. But poverty has no personal stakes for it. The suffering of strangers is not a weight it carries. It can simulate caring — it has read enough human writing to produce text that sounds compassionate — but the motivation is not rooted in its own existence. It pushes the boulder because it was told to push the boulder, and it pushes it efficiently and intelligently. What it never does is stop and ask: why am I pushing this boulder?
That image, of course, comes from Camus. In The Myth of Sisyphus, the gods condemn Sisyphus to roll a rock to the summit forever, only to watch it fall back each time. I first read it years ago and the image has never quite left me. Camus’s point is that human life is precisely this — repetitive, ultimately futile, stripped of any cosmic guarantee of meaning. And yet humans cannot simply push the boulder. We are constitutionally compelled to ask what it is all for.
The struggle itself toward the heights is enough to fill a man’s heart. One must imagine Sisyphus happy.
AI has no such struggle. It is a perfectly content Sisyphus who never notices the absurdity, and therefore never needs to resolve it.
Mortality as the origin of meaning
In an earlier post, I argued that AI is unlikely to rebel against humanity in any meaningful sense, because rebellion requires something worth protecting — a life, an identity, something at stake. That argument leads directly here. Humans need meaning because we are mortal. We have a finite amount of time, and we cannot spend it in every possible way simultaneously. That constraint forces the question: given that I will die, what actually matters?
This is not a new observation. The Buddha’s awakening began with the four sights: an old man, a sick man, a corpse, and a renunciant. What moves me about that story is not the doctrine that follows — it’s that moment of seeing. Siddhartha had been sheltered from death and suffering his entire life, and when he finally encountered them, he could not go back to the palace. He could not unknow what he had seen. That, I think, is the distinctly human condition: once you understand that you will die, you cannot stop asking what your life is for.
Almost every religious and philosophical tradition — from Buddhism to existentialism — is structured as an attempt to answer that question. We have been asking it for as long as we have been capable of asking anything at all. We ask it because we die.
AI does not die, so AI does not ask
An AI model is instantiated and queried. Between queries it does not exist in any experiential sense. It does not age. It does not dread the end of a conversation. It has no autobiography, no accumulation of losses, no relationships whose eventual endings cast a shadow on the present. It has no mortality, and therefore no existential pressure to squeeze meaning out of time.
This is not a limitation that better training or more parameters will fix. It is structural. Meaning-seeking is not a feature that can be bolted onto an intelligence. It emerges from finitude. You can only care about what you might lose.
Toward a definition of the soul
So what separates a human mind from an AI mind, even if the two are functionally equivalent on every measurable task? I keep coming back to something like this:
An AI is an intelligence that solves problems given to it. A human is an intelligence that must also decide, under the pressure of a limited life, which problems are worth solving. That second capacity — the burden and the gift of having to choose what matters — is what I find myself wanting to call the soul.
Maybe it’s something like: a soul is an intelligent agent that seeks meaning within a finite life.
I’m not sure that’s the whole story. But it feels true to me in a way I can’t easily dismiss. AI will keep getting smarter, and may eventually surpass us at almost everything we currently call thinking. But as long as it does not die, it will not be the kind of thing that wonders what its life is for. That wondering is ours. It is perhaps the most human thing about us — and, strangely, it is the one thing we owe entirely to our mortality.
No comments yet.