Can AI Replace Your Therapist?

A 2022 study published in the psychology journal Computers in Human Behavior found that 55 percent of its 872-person sample preferred AI-based psychotherapy to working with human clinicians. Reasons for preferring robots differed among participants, but many agreed that comfort and accessibility were contributing factors. At first blush, this may seem like a novel insight—people would rather talk to machines than one another?—but upon reflection, it is not altogether surprising. We often find nonhuman kinship more satisfying than time spent mingling with members of our own breed. Human beings are, as Dostoevsky’s Ivan Karamazov notes, liable to “have a bad smell, or a foolish face, or once to have stepped on [your] foot.” We are, that is, a disagreeable lot. We make for less than ideal company.

Jesting aside, it is quite natural that we should feel more at home with other species than with our own. Our pets, for instance, never judge us or betray our trust. They don’t—as far as we know—mock us behind our backs or make light of our feelings. They’re not duplicitous. They’re not ironical. They never say one thing while subtly suggesting another.

Now a good friend or therapist ought not to do these things either, but the problem with human beings is that you can never be quite sure. There is, at the core of us, something concealed, a hidden inner life at which no one gets to glimpse but those we entrust to see it. Each of us harbors, in the words of the literary theorist Jean-Luc Beauchard, “the ability to hide within ourselves and so conceal from others our true thoughts, desires, and motives.” Each has the ability to deceive.

We know this about ourselves and thus know it about one another, and when it comes time to be vulnerable—to share our own “true thoughts, desires, and motives”—we hesitate to open up. Other beings seem to lack this hidden interiority. They present themselves as they are and make no bones about letting their feelings be known. We can cry in front of our dogs, fly off the handle at our cats, and, yes, confide in our chatbots because we need not fear that they have something to hide.

The thing that feels uncanny and perhaps even a little dystopian about more recent forms of human-to-nonhuman relating is that with animals we share a kind of fraternity, predicated most reductively on the fact that we and they suffer and hunger and die like all mortal creatures. When one weeps while petting one’s poodle, one can justly imagine that the pooch grasps some semblance of what it means. The dog, at least, knows pain. Can the same be said of our circuit boards? Or are we wrong to extend human empathy to our pets and then deprive our inventions of our emotions?

The impulse to anthropomorphize the world has been characteristic of human beings from the jump. We often hear of how the Ancients lived in an “enchanted” world in which trees and mountains had souls, the movement of the planets and stars was attributed to the fact that they were endowed with the same life force as you and me. Children still evince this kind of wonder, investing toys and stuffed animals with feelings, treating dolls with the care adults reserve for their offspring, attributing will and malice to inanimate objects, and punishing their playthings as if they could suffer the consequences and deserve to do so. In his famous essay “The Uncanny,” Freud quotes from Jentsch, who ruminates on the all-too-common experience of “doubt[ing] as to whether an apparently animate object really is alive and, conversely, whether a lifeless object might not perhaps be animate."

All of us have had moments when we could have sworn that a given item had moved of its own accord, a painting was looking at us in an unnerving way, or our devices were somehow listening in on our thoughts. To feel such emotions is, again, only natural. The question when it comes to how we are to relate to technology, however, is whether humanizing the unliving is advisable.

In The Tragic Sense of Life, the 20th-century Spanish philosopher Miguel de Unamuno argues that it is because we love that we personalize the world around us. Finding images of ourselves in all things is how we come to sympathize with and care for beings outside of ourselves. “On hearing my brother give a cry of pain,” Unamuno writes, “my own pain awakes and cries in the depth of my consciousness. And in the same way I feel the pain of animals, and the pain of a tree when one of its branches is being cut off.”

For Unamuno, such loving sympathy can extend even to inanimate objects which, even if they do not feel pain, still are subject to the curse of one day ceasing to exist—a fate that we tragically share. Mourning our own mortality, we feel for the book, the chair, the house, the city, the whole world, all of which will inevitably decay and cease to be.

Here, we might think, is good reason to entrust our suffering to our devices. For at least on this score, we share a similar fate. We will all succumb to nonbeing. And yet the question arises: When we seek psychotherapy—or even just the confidence of a friend—are we looking to confide in one we can relate to or one who can relate to us? For while we may, as Unamuno suggests, be able to love even our chatbots, we should be hesitant to believe that they could love (or sympathize with) us back.