Deagreez/iStockphoto/Getty Images Myra Cheng, a computer science Ph.D. student at Stanford University, has spent a lot of time listening to undergraduates on campus. “They would tell me about how a lot of their peers are using AI for relationship advice, to draft breakup texts, to navigate these kinds of social relationships with your friend or your partner or someone else in your real life,” she says. Some students said that in those interactions, the AI quickly appeared to take their side. “And I think more broadly,” says Cheng, “if you use AI for writing some sort of code or even editing any sort of writing, it’ll be like, ‘Wow, your code or your writing is amazing.’ “ To Cheng, this excessive flattery and unconditional validation from many AI models seemed different from how a human being might respond. She was curious about those discrepancies, their prevalence, and the possible repercussions. “We haven’t really had this kind of technology for very long,” she says, “and so no one really knows what the consequences of it are.” …