When We Treat Robots Like Friends: Untangling the "Human-Like" Mess
Imagine you are walking through your house and you accidentally trip over your small, round vacuum robot.
"Ouch! Sorry, little guy!" you might say.
In that moment, you treated a plastic machine like a living friend with feelings. This is a huge deal for scientists because, as robots move into our schools and hospitals, we need to know why we treat them like humans—and who is responsible when they "trick" us.
The Problem: A Big Mess of Definitions
A Literature Review Reveals Chaos
Researchers just looked at 57 studies from the years 2000 to 2020 and found a big mess. It turns out scientists were using 7 different definitions for "human-likeness."
The New Solution: Two Words, Not One
To fix this, a new study from the University of Cambridge says we need to stop mixing up what the robot is with how it makes us feel. They say we need two different words for two different things.
Anthropomorphism
This is like a "brain trick" where you decide a robot has a personality, even if it’s just a toaster with googly eyes.
Anthropomimesis
This is like a "robot costume" where an engineer specifically builds a robot to look, move, or think like a person.
M.
Axelsson
In our definition of anthropomorphism, imbuing is in the eye of the beholder, i.e., the user/perceiver of the robot... we hold this to be distinct from anthropomimesis, in which human-like qualities are designed into the robot by its developer/designer.
Building the "Robot Costume"
The scientists say there are three ways a designer can build this "robot costume."
Aesthetic
Make it look human.
Behavioural
Make it act social.
Substantive
Build it with fake muscles and joints.
The Spectrum of Mimesis: From Weak to Robust
"Weak" Mimesis
Some robots have "weak" mimesis, like the old computer program ELIZA that just repeated words back to you.
"Robust" Mimesis
Others, like the newest robots powered by Large Language Models—which are like giant "word-guessing engines"—have "robust" mimesis because they are very good at acting human.
Why This Distinction Matters
Key Takeaway: If a robot breaks a rule or hurts someone's feelings, we need to know if the user was just imagining things (anthropomorphism), or if the designer built a machine meant to manipulate them (anthropomimesis).
The Path Ahead: What We Still Need to Learn
Unanswered Questions
There is still a lot to learn, though. The researchers haven't tested these new words in a real lab with kids or adults yet.
They also haven't looked at robots that act like dogs or cats.
Final Thought: For now, they want everyone to remember: a robot's "soul" is something we give it, but its "skin" is something a scientist built.
Source: Axelsson, M., & Shevlin, H. (2026). Disambiguating Anthropomorphism and Anthropomimesis in Human-Robot Interaction. University of Cambridge, UK. arXiv:2602.09287v1 [cs.RO].