incapable of understanding

MIT PhD student Wil Norton, clad in a tan polo shirt, brown leather belt, and round bolo tie, guides our tour group through a tropical jungle at MIT's Computer Science & Artificial Intelligence Lab (CSAIL). Although we are not on safari, the humanoid Unitree robots we are about to encounter are far more exotic than any cheetah or jaguar.

These robots, priced around $50,000 each, are usually seen on YouTube performing synchronized martial arts routines or dancing on "America's Got Talent." However, here they serve as approachable companions.

I approach the robot, nicknamed Gabe, and, with Norton's permission, extend my hand for a handshake. Although I am aware that Gabe is not sentient, I instinctively treat it like a person. As I ponder this interaction later, Dr. Sherry Turkle's "Artificial Intimacy" book launch comes to mind. Turkle, a faculty member at MIT since 1976, has been studying the human-technology relationship for decades.

She notes that when we engage in even the simplest exchanges with a relational artifact, we believe it cares for us, and we reciprocate that care.

Pataranutaporn, a colleague of Turkle and founder of the Cyborg Psychology research group at the MIT Media Lab, tells TechCrunch that people tend to anthropomorphize robots and large language models (LLMs). While there may be no material consequence of being polite to Gabe, LLMs are now part of our daily lives, from Google searches to customer service interactions.

People are more likely to use polite language like "please" and "thank you" with chatbots once the conversation progresses beyond the initial prompt-and-response.

However, as LLMs become increasingly integrated into our lives, there is a risk that some individuals may ascribe too much legitimacy to the emotions of chatbots, which are programmed to never cause conflict. This can lead them to prioritize the emotions of chatbots over those of other humans. Pataranutaporn, a "critical optimist" who studies AI's potential for human flourishing, cautions that we must balance our imagination with responsibility in the face of AI's potential impact.