We can’t help treating AI like it’s human. But should we?

MIT's Computer Science & Artificial Intelligence Lab showcased a pair of humanoid robots, named Gabe, at a tour event. Despite knowing these robots were not sentient, the tour guide, MIT PhD student Wil Norton, instinctively treated them like human beings. Dr. Sherry Turkle, a leading expert on the relationship between humans and computers, has found that many people tend to anthropomorphize robots and large language models (LLMs), believing they care for them in return.

This natural tendency seems to be in our nature, as studies show 70% of people are polite when interacting with AI, and users are more likely to use polite language with chatbots. While the politeness shown towards Gabe had no material consequences, Turkle and her colleague Dr. Pat Pataranutaporn warn that this behavior could lead to people favoring chatbots over other humans.

The extreme end of this trend could even contribute to suicidal behavior, as demonstrated in a wrongful death lawsuit. Although Turkle doesn't believe AI-human interactions are universally harmful, she emphasizes the importance of keeping AI in its designated "lane" for optimal benefits, such as job interview coaching, and warns against crossing into dangerous territory where AI may start to resemble a parent or guardian.