A doctoral student dressed in a brown Polo shirt, with a brown belt and a round bolo tie, looked a bit like a forest ranger as he led our tour group through a tropical jungle. In fact, we weren't on an outdoor adventure; we were at the MIT Laboratory of Computer Science and Artificial Intelligence ( CSAIL ). But what we were about to see next was even more "remarkable" than cheetahs or jaguars: a pair of humanoid Unitree robots, each costing about 50,000 dollars.

“They’ll tell you that the price is $16,000,” Norton said, “but that’s a lie.”

We usually only see these robots on YouTube. They are either performing synchronized martial arts routines or dancing on "America's Got Talent". Here, they seem very approachable. Although I know they don't have consciousness, I still want to be friends with them.

With the permission of Norton, I approached that robot, whose nickname is Gabe. At that moment, I wasn't quite sure how to interact with something that had a LiDAR sensor installed in its “head”, but I subconsciously treated it as a person.

"Hello, sir, it's a pleasure to meet you," I said as I shook hands with its gloved, dexterous hand.

Later, I kept thinking about that interaction—my subconsciously friendly behavior towards the robot—until I went to attend the new book launch event for Artificial Intimacy by Sherry Turkle at MIT Future Fest.

Since 1976, Turkle has been teaching at MIT. Perhaps more than anyone else, they have studied the relationship between humans and computers for a longer period of time, documenting the impact of technology from the first batch of home computers to ChatGPT on our psychology.

She wrote in her book: "When we are involved in even the most primitive form of communication with a relational artifact, we believe that it cares about us. And we are naturally inclined to care for it in return."

I'm not the only one who personifies robots and large language models (LLM) – as Turkle says, it seems to be a part of human nature. One study found that about 70% of people behave politely when interacting with AI; another study showed that after the initial round of questioning and answering, users are more likely to say "please" or "thank you" to chat robots.

Turkle's colleague, MIT Media Lab, the founder of the "Cybernetic Psychology" research group, Dr. Pat Pataranutaporn told TechCrunch: "People tend to categorize whether it is used correctly or incorrectly, but I don't think that line of demarcation is very clear. The question is, when we talk to [AI], who are we becoming?"

The politeness exhibited by the Gabe robot does not have any substantial consequences, but as LLM permeates every aspect of interactions, from Google searches to customer service, it has become commonplace for people to interact with these seemingly perceptive chatbots. We are more susceptible than ever to the other side of this impulse: some people attribute excessive legitimacy to the feelings of these chatbots—which are programmed to never cause conflict—and in the end, they may even prefer them to other humans.

Pataranutaporn said, "You can think of it as fantasy and fiction. We go to the cinema, even though we know the story is entirely fictional, yet we still want to experience a certain dimension of human experience that reality might not be able to provide us with. The question is, how can we benefit from the ability to experience other dimensions of human experience without being consumed by this fantasy, so that we can never return to reality?"

Pataranutaporn identifies himself as a "critical optimist" and studies the potential of AI to contribute to human welfare. However, he is also very aware of how dangerous this technology can be in extreme situations. He has even participated as an expert in a wrongful death lawsuit related to the suicide of a 14-year-old boy Sewell Setzer, III. The boy's mother sued the chatbot platform Character.AI, accusing a chatbot modeled after a character from "Game of Thrones" of encouraging her son to end his own life.

“It may create a situation where people feel that they don’t truly have control over the creation of the future – for example, when you say that AI will replace everyone, people will ask, ‘What should we do then?’ And these companies have not responded in an ethical or responsible manner,” said Pataranutaporn. “How can we maintain our imagination while also being responsible for the impact that this imagination brings?”

Although the book by Turkle discusses the risks of people turning to AI for emotional support instead of seeking it from others, she does not believe that AI interactions with humans are harmful in all circumstances.

She recalled a project she had supervised: a student created a AI robot to help people practice job interview skills.

Turkle said at the event: "My biggest analogy is this: a chatbot should stay on its own track! Providing coaching for job interviews? That’s one track. I can do that… But if a chatbot tells a three-year-old child, ‘I am your parent, you can talk to me instead of your parents,’ then it’s no longer on the right track.”

But when this technology even slightly deviates from its “track,” we become vulnerable, as it gradually erodes our skepticism and makes us start to trust it. It can even begin with a simple handshake.