A new pair of studies from MIT Media Lab and OpenAI found that those who use the chatbot most heavily also experience the most loneliness. The catch-22: it’s unclear whether this is caused by the chatbot itself or if lonely individuals are simply more likely to seek out emotional bonds. Researchers analyzed millions of interactions and found that only a small number of users rely on ChatGPT for emotional support—but those who do are among its heaviest users. The MIT study found that higher daily usage of ChatGPT “correlated with higher loneliness, dependence, and problematic use, and lower socialization.” Since loneliness is a tricky feeling to quantify, researchers said they measured both users’ subjective feelings of loneliness and their actual levels of socialization. The studies also found that heavy users were more likely to consider the chatbot a “friend” or attribute human-like emotions to it. Those engaging in “personal” conversations with the chatbot reported the highest levels of loneliness. If they set the chatbot’s voice mode to the opposite gender, those levels were even higher. It’s been over two years since OpenAI released ChatGPT. While researchers emphasize that these studies are preliminary, they reinforce existing concerns about how AI chatbot tools are affecting people’s lives—both online and offline. ChatGPT attracts 400 million users weekly worldwide. Some use it to win arguments or even as a substitute for therapy, despite warnings from health professionals. Others call ChatGPT their “best friend.” “Interactions with chatbots that cater to your preferences and are trained to be as polite and affirming as possible might help in the moment when you interact with them, but they also slowly chip away from your ability to deal with the messy real world and complex human interactions,” says Sandra Matz, Columbia Business School professor and author of MINDMASTERS: The Data-Driven Science of Predicting and Changing Human Behavior. “The problem is that we’d need to understand this more causally by assigning people to use or not use chatbots and then studying the impact of these experiences on loneliness,” she adds. “Obviously, something that comes with ethical questions if we’re playing around with people’s experience of loneliness.” There’s been increasing scrutiny of the negative effects of interacting with AI chatbots—and for good reason. Decades later, researchers are still trying to fully grasp the impact social media has had on mental health. When it comes to AI chatbots—well, I guess we’ll check in again in a couple of years and see.