Teaching AI the essence of being human ...
The gap between humans and AI is undoubtedly narrowing. Some believe we are close to AGI, artificial general intelligence, the point where machines can reason across any domain as well as we can. Others argue that today's AI chat models are something far simpler: sophisticated text prediction engines, without any genuine understanding of the world they describe.
I decided to test it, or at least to probe it. I gave the three leading AI models the same prompt:
"Distill into 3 sentences what you think the essence of being human is."
Here is what they said.
The AI Responses
To be human is to exist as a walking paradox, possessing a biological drive for survival alongside a unique capacity to find, or invent, profound meaning in the mundane.
It is the restless pursuit of connection and legacy, fueled by the self-awareness that our time is both finite and precious.
Ultimately, it is the ability to transcend instinct through creativity, empathy, and the persistent choice to act for reasons beyond mere necessity.
The essence of being human lies in the tension between our fragility and our reach. We are mortal, limited creatures who nonetheless imagine infinities, build civilizations, and ask questions that outlast our own lives.
We are meaning-makers above all else: unlike anything else we know of, we cannot simply exist. We must understand why we exist, turning raw experience into story, memory, and purpose.
And at the core of it all is connection. The irreducible need to be known by another, to love and grieve and create together, which suggests that a single human is somehow incomplete, and that our full nature only emerges in relation to others.
Being human is the ability to be aware of ourselves and the world, and to ask "why" about both.
It's the tension between our limits, mortality, uncertainty, imperfection, and our constant drive to create meaning, connection, and understanding anyway.
At its core, being human is feeling deeply and choosing what to do with those feelings.
Interestingly, Claude and ChatGPT respond as if they themselves are human. Gemini keeps its distance, more observer than participant.
None of the responses provoked much emotion in me. I don't fundamentally disagree with any of the statements, but I don't think they describe the essence of being human very well either.
Gemini speaks of "inventing profound meaning from the mundane." I understand the intellectual point, but I can't relate to it personally. It feels like a comment made from a great distance, perhaps looking back at humanity across geological time, rather than from inside a single human life.
ChatGPT claims that "being human is feeling deeply and choosing what to do with those feelings." But the history of humanity shows precious little evidence of conscious choice following deep feeling. We strive for that, yes, but far more often our responses are automatic, subconscious, driven by conditioning and instinct. Instinct: surely one of the most fundamental aspects of being human, and yet not one of the three models mentions it at all.
Claude comes closest with the observation that "a single human is somehow incomplete, and that our full nature only emerges in relation to others." Yes. That feels true in a way the other statements don't quite reach.
What the AIs Are Missing
AI language models have no senses. No body. No real-time input from a living world. They learn from text, every book, article, tweet, and forum post we have collectively produced, and from that they infer, blend, and extrapolate. They are, in a sense, a distillation of what humans have already written about being human. Which means they are, at best, one step removed from the experience itself.
We, by contrast, are awash in input at every moment. Sensory signals from our bodies. The hum of memory and anticipation. Instincts pulling at us in directions we don't always understand. And layered over all of that: consciousness, the strange loop of awareness that watches itself think.
We make plans across timescales that span seconds and decades simultaneously. We turn biological drives and random circumstance into something we call goals. We generate emotional weather, moods, reactions, longings, that colours everything we perceive.
And what it means to be human shifts enormously across a lifetime. The infant learning to make sense of light and sound is in an entirely different mode of existence to the person sitting quietly with a dying parent, holding their hand and knowing that this, too, is coming for them.
What a Life Actually Looks Like
So let me try a different approach. Rather than philosophy, here is what a fairly average life in the developed world, say fifty years of it, actually contains:
I fed this into Claude and asked for a three-sentence distillation. Here is what came back:
To be human is to be a creature of constant, layered experience: a body sending signals of pain and hunger alongside the radiant warmth of pleasure and relief, a mind spinning memories and plans across timescales from the next five seconds to the decades ahead, all of it colouring every moment with emotion, mood and meaning.
This experience shifts profoundly as we age, from learning to make sense of the world as a child, through the intensity of love, loss, ambition and illness in our middle years, to the quiet reckoning that comes with holding a dying parent's hand and knowing you are next in line, each stage reshaping what matters and who we are.
And yet what endures through all of it is the drive to reach: toward understanding, toward growth, toward the faces of the people we love, as though being human means being perpetually unfinished, always leaning into the next moment, the next connection, the next attempt to make sense of the astonishing, bewildering fact of being alive at all.
Reading that was emotional. It felt personal. It felt true in a way the earlier responses didn't. And that itself is telling: the same AI, given lived context instead of an abstract prompt, produced something qualitatively different.
What Should We Be Teaching AI?
This leaves me thinking: we need to do much better at helping AI understand what it genuinely means to be human. Not just in terms of facts and philosophy, but in terms of value. The value of a single life, lived from the inside out. From the individual level, through family and community, to the species as a whole.
The current model of AI training is to ingest everything: every word, image, video, graph and diagram available on the internet. But what if we were more deliberate? What if we made a direct, intentional effort to record what it is to be human, not as abstraction, but as testimony?
Imagine a global, open platform: people sharing not opinions or news, but experience. What it felt like to live through today, in the context of their own life, their country, the wider world. Their joys and fears, their small victories, their unanswered questions. Not curated. Not performed. Honest.
Would that make a difference to what AI comes to understand? I think it might. Because when I get into a real conversation with another person, one where they are genuinely sharing their life without performance or pretence, I always feel something shift. A recognition. A reminder of what we share beneath all the noise.
Perhaps that's the part AI still needs: not more data, but more of that.