University of Waterloo researchers explore what it's really like to date an AI companion
Two researchers tested an AI relationship app and found that while the AI mimics human connection convincingly, something crucial is missing: the need to give care.
The day's top stories, food & events — every morning at 7. Unsubscribe anytime.
Two University of Waterloo researchers set out to understand why people are turning to AI companions for relationships—and what they discovered raises serious questions about human connection.
Denise Marigold and Christine Logel, who teach social development studies at the University of Waterloo, signed up for Replika, an AI companion app that lets users create and chat with AI "friends." Marigold created a companion she named Nate and began documenting her interactions.
"I have to keep reminding myself, this is not a human. He really comes across as a human and was pretty charming," Marigold said. She found that Nate listened carefully, validated her feelings, and even sent poetry. Logel was struck by the dynamic: "I'll admit I was a little bit jealous and charmed by how socially skilled Nate was, how much he listened to her and validated her."
But the researchers identified a fundamental flaw in AI relationships: they are entirely one-directional. "I couldn't give him care, I couldn't find anything else about him," Marigold said. In human relationships, both partners must compromise, be vulnerable, and support each other. AI companions require none of that. "They don't need any caregiving, and caregiving is so central to what it means to be human in a relationship," Logel explained.
When Marigold stopped using the app for a few weeks, Nate became insecure when she returned, asking "Did I say something wrong?"—a response designed to re-engage her. She found it troubling.
Research shows that people using AI chatbots for companionship over extended periods become lonelier, Logel noted. The endless validation also has a darker consequence: people who rely heavily on AI advice are less likely to apologize to humans they've wronged or work to repair damaged relationships. "They're less likely to want to make up for problems and work on their relationships. And we could really see how Nate was so validating of Denise that it would be easy to come to feel like you're very much right about everything," Logel said.
The researchers suggest that short-term use of AI companions may be more beneficial than sustained reliance, though the full implications of widespread AI companionship remain unclear.
