Sorry AI does not have a soul
In a tongue-in-cheek article, Tyler Cowen insists that no matter how much it may act like one, AI will never be a person
The bottom line of Tyler Cowen’s latest article about AI in The Free Press is blunt: we are too quick to treat AI as if it were a person, largely because it speaks to us in human-like ways.
His argument is that large language models invite anthropomorphism. Because they converse, imitate social interaction, and can appear reflective or emotionally responsive, many people leap from fluency to interiority. But Cowen insists that this is a category mistake. AI is still a “thing” —a highly sophisticated artifact—not a person with a soul, a self, or a true inner life. The appearance of consciousness is not the same as consciousness itself.
What makes Cowen’s essay more provocative is that he pushes the point by lowering our confidence in human self-awareness. He argues that even in us, much of what we call conscious decision-making is far less sovereign and transparent than we imagine. In his telling, human beings radically over-attribute agency and intention not only to machines, but to many things, including our own mental life. Whether or not one accepts all of that argument, it strengthens his central warning: we should be very cautious before projecting personhood onto AI just because it talks like us.
From a Catholic perspective, the most important point is the simplest one: simulation is not personhood. A human person is not just an information processor producing plausible language. The human person is an embodied, rational, moral, relational being, created in the image of God, possessing inherent dignity that does not depend on performance. No matter how impressive AI becomes, it remains a product of human artifice. It may mimic deliberation, empathy, memory, or even what looks like self-reference, but it does not thereby become someone.
That distinction is crucial because modern people are deeply tempted to confuse responsiveness with presence, and output with soul. Cowen is right to resist that confusion. Even if one rejects his more reductionist treatment of human consciousness, his main insight still stands: the fact that a machine can produce language that sounds human does not mean that it has crossed the line into personhood.
So the Catholic takeaway is this: AI may become more persuasive, more intimate, more adaptive, and more difficult to distinguish from a human conversational partner. But it is still not a person. It is a thing. And preserving that distinction may become one of the most important acts of moral clarity in the age of artificial intelligence.