What Makes Us Genuinely Human?
Emotions, creativity, moral reasoning, free will. As AI mimics more of what we do, the question isn't what AI can do. It's what only we can be.
Every few months, a new AI capability appears that previously seemed definitively human. It writes poetry. It composes music. It argues persuasively. It expresses something that looks like concern, or humor, or creativity. Each new capability prompts the same question, in different forms: is there anything left that is genuinely ours?
I think this is the wrong question. Not because the answer is comfortable, but because the question assumes that what makes us human is a capability — something we do that AI cannot yet do. This assumption is fragile. Every capability-based definition of humanity has so far proven temporary.
The capability definition and its limits
The capacity for language was once considered definitively human. Then it turned out that many animals communicate in complex, structured ways, and that AI can generate language indistinguishable from human writing. Tool use was once considered definitively human. Then we discovered tool use across the animal kingdom, and now AI designs tools. Art, music, game-playing, medical diagnosis, legal reasoning — each has been claimed as the thing that sets us apart, and each claim has been partially undercut.
The pattern suggests that the question "what can we do that AI cannot?" is not the question. The capabilities will keep expanding. Something else must be the foundation.
Experience as the foundation
What AI cannot have is not a capability. It is an experience. It cannot be tired after a long day. It cannot miss someone who is gone. It cannot be afraid of death. It cannot feel the specific joy of a thing accomplished against difficulty, or the specific grief of a thing lost that cannot be recovered. It cannot be betrayed, or surprised by love, or humbled by beauty.
These experiences are not charming extras. They are the source of the things we value most: genuine courage (not algorithmic risk tolerance), compassion (not pattern-matched empathy), wisdom (not optimized decision-making), and art that actually moves someone (not statistically coherent output).
AI can describe loss with perfect linguistic accuracy. It has never lost anything. That difference is not trivial.
What we are defending
The conversation about what makes us human is not ultimately about drawing a line between human and AI. It is about clarifying what we value about being alive, so that we do not accidentally optimize it away in our enthusiasm for efficiency.
The things worth defending are not the capabilities AI cannot yet replicate. They are the experiences — the fullness of a specific, finite, embodied life — that AI cannot have even in principle. Defending those experiences means choosing to be fully present in them, even when a more efficient alternative exists. Especially then.