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

In a tan polo shirt, brown leather belt, and round bolo tie, MIT PhD student Wil Norton resembles a park ranger leading a tour group through a tropical jungle. However, this isn’t a safari; we are at MIT’s Computer Science & Artificial Intelligence Lab (CSAIL), where we encounter something even more exotic than wildlife: a pair of humanoid Unitree robots, each costing about $50,000. Norton humorously notes, "They’ll tell you they cost $16,000. That’s a lie."

These robots are typically seen on platforms like YouTube, performing synchronized martial arts routines or dancing on shows like America’s Got Talent. Here, they are approachable, and despite knowing they are not sentient, I feel an instinctive desire to connect. With Norton’s permission, I approach the robot nicknamed Gabe. Unsure how to interact with a machine equipped with a LiDAR sensor instead of a brain, I default to treating it like a person. "Hello, sir, it’s nice to meet you," I say, shaking its gloved, dextrous hand.

Reflecting on this interaction later, I attend Dr. Sherry Turkle’s book launch for "Artificial Intimacy" at MIT Future Fest. Turkle, a faculty member at MIT since 1976, has extensively researched the relationship between humans and computers, examining technology’s psychological impact from the era of home computers to the age of ChatGPT.