Feeling Is The Human Superpower AI May Never Learn
Regan Peng is an AI founder and former senior tech executive, focused on personal AI, privacy-preserving systems, and edge AI.
gettyImagine an old voicemail from someone you have lost. You know every word. You listen anyway.
An AI might eventually explain every mechanism behind that attachment. It could recognize the voice, reconstruct the relationship and predict your response. None of that would, by itself, establish that it knows what it is to miss someone.
Humans are feeling machines that can think. AIs are thinking machines that try to feel.
The second half describes an engineering aspiration, not an established inner life. Intelligence and subjective experience are not interchangeable concepts.
My argument is that feeling is our deepest power: the capacity for existence to matter to us. It should never become something we apologize for as machines grow more capable.
The human half of that contrast owes a debt to neuroscientist Antonio Damasio. His account locates feelings in the relationship between the nervous system and the body it helps regulate. Hunger, pain, comfort and attachment belong to a biological history shaped by evolution. They sustain life, rather than merely decorating our thoughts.
Millions of years stand behind this inheritance. Yet it takes on a particular history in each of us: This place becomes home; this person becomes irreplaceable. I see feeling as one of evolution’s most extraordinary achievements—life becoming significant to itself.
A recent philosophical account of consciousness develops a related argument: Experience may depend on being a subject for whom things can genuinely go well or badly. On this view, processing information is insufficient without the internal organization that makes outcomes matter to the system itself.
This changes how we might regard our evolutionary origins. Explaining love through the history of survival and reproduction does not establish that love is merely a reproductive strategy. An account of where something comes from cannot, on its own, decide what it is worth.
Nor does calling feeling a superpower mean our feelings are always right. Fear can imprison us; affection can distort judgment. Reason helps us examine both. The argument is against confusing freedom from error with freedom from caring.
In a September 2026 paper, Geoffrey Hinton and his coauthors examine how automating AI research could create an accelerating cycle: Capable systems help build still more capable successors. An intelligence explosion is a possibility they take seriously, not an established outcome.
Even if it happens, the question of feeling survives. Greater ability to explain the world does not, by itself, establish that a machine experiences being in it.
Physical AI sharpens the distinction. Robotic touch sensors can register pressure and heat. Whether holding someone’s hand brings comfort is another question.
Some researchers argue that consciousness may depend on properties of living systems, rather than computation alone. Others allow that artificial systems could acquire the organization and relationships needed for experience. The issue remains unsettled.
AI may never learn to feel simply by becoming better at thinking. We should take that possibility seriously instead of assuming that an inner life arrives automatically with sufficient scale.
And should machines eventually feel, our feelings would not become obsolete. Another being’s capacity to love does not diminish yours. Human worth should never have depended on possessing a capability nobody else could acquire.
Perhaps feeling is a kind of password to an interior dimension of existence. I mean this as a philosophical image, not a claim about hidden physics: Through feeling, the universe becomes something we inhabit intimately.
We need not invoke another world to recognize how extraordinary that is. Within this one, we can be moved by a voice, overwhelmed by beauty or changed by someone else’s suffering.
Consider time. An hour waiting for someone carries something a clock cannot record: hope that they will arrive. Perhaps part of our wish for immortality comes from wanting that encounter with existence to continue. We still want to feel more.
For those of us building AI, this changes what progress should mean.
A tool that saves an hour and immediately fills it with more work has not necessarily enriched a life. An assistant that recognizes loneliness should help strengthen a person’s relationships, not make the product harder to leave. Emotional understanding should serve the person who feels, rather than make that person easier to influence.
The point of building more capable machines should be to give people more room to live.
That asks something of us, too. We should resist using intelligence to insulate ourselves from every uncertainty. Preventable danger deserves safeguards. But a relationship whose future is unwritten, an unfamiliar idea or a discovery that unsettles us is not merely a prediction problem waiting to be solved.
Curiosity means remaining available to what we have not yet experienced. We can welcome a universe that becomes more intelligible without demanding that it stop surprising us.
Perhaps this is what each generation can leave the next: better ways to understand the world, and more freedom to encounter it. Future artificial minds may join that inheritance. They need not become the standard against which every human life is judged.
We do not have to remain the most intelligent beings imaginable for our lives to matter.
The voicemail is not precious because we cannot explain it. It is precious because we still want to hear the person’s voice.
Forbes Technology is an invitation-only community for world-class CIOs, CTOs and technology executives. Do I qualify?
