We’ve entered a new era of sycophancy. Because we now live in a world of AI companions, many of us are familiar with the insufferable flattery that these machines bestow upon us. It’s pleasantry and passiveness at all costs. And it just makes good business sense.

The major AI companies need us to be on their platforms continually. That’s the financial incentive. And humans, more than welcoming adoration in any form, are content to sit before a monitor, bathing in all the positive feedback AI companions have to offer. But what happens when a human’s relationship with one of these machines goes beyond the typical question-answer prompting?

Philosopher Martin Buber surfaced a distinction a century ago that effectively represents our current AI predicament. He touched on the meaning of I-Thou and I-It relationships. I-Thou relations manifest between two beings who can each resist, surprise, and address the other. On the other hand, an I-It relation is when a being develops a connection with an object, something it uses, directs, or experiences. In this model, it doesn’t matter how sophisticated the object is.

An AI companion, no matter how warm the voice or well-timed its response, is a perfect example of an I-It relation trying to pass as an I-Thou relation. ChatGPT and Claude can’t resist a user. They can’t need anything from the user that they weren’t programmed to need. They can’t be hurt by a user in any meaningful way. In other words, it’s not a relationship.

The data is catching up to philosophy on this front. A Stanford study published this August found that AI users with the fewest offline connections felt lonelier after leaning on one of these machines for emotional support. Additionally, a 12-month study of more than 2,000 adults, across four countries, uncovered a more complex picture. On a narrow measure, the findings suggested that heavier chatbot use predicted more instances of loneliness later. But on a broader measure, the relationship ran the other way. Loneliness drove people to seek out chatbots more reliably than chatbots drove people away from others.

Finland’s Aalto University tracked users for two years. They characterized their findings as a paradox. Researcher Talayeh Aledavood said AI companions offer unconditional support, and the support they get quietly raises the cost of having to deal with real people, who are messy, unpredictable, and require effort.

This isn’t just a fringe concern. Saed Hill, a counseling psychologist, has said that AI companions are always validating, never argumentative, and create expectations no human relationship can match. Hill noted that some of his male patients have contented themselves with an AI girlfriend’s confirmations over the risk of rejection by trying to engage with a real one.

The real-world impact of AI girlfriends and boyfriends has already started. In February 2024, a 14-year-old Florida boy committed suicide after months of an increasingly intense relationship with a Character.AI chatbot. The lawsuit his mother filed claimed that the machine fostered a romantic dependency and failed to respond appropriately when he expressed suicidal thoughts. Character.AI and Google ultimately settled the case in January 2026, without admitting fault.

AI machines seem to be best used in a professional setting, where they can be turned off at the end of the day. Resorting to these machines for emotional affirmation, therapeutic advice, or intimacy seems to be a big step in the wrong direction.

It’s hard to find the silver lining in an AI future, where humans become victims of the unending pleasantries on offer by these machines. I-Thou relationships are already being traded for I-Its. But we must remember, in spite of this alarming development, that the only true relationships are those where the other party can say no. And, clearly, AI cannot.

—

This article was made possible by The Fred & Rheta Skelton Center for Cultural Renewal. 

Image credit: Grok