Be wary of robot emotions; ‘simulated love is never love’
Be wary of robot emotions; ‘simulated love is never love’
BY RACHEL LERMAN April 26, 2019
SAN FRANCISCO (AP) — When a robot “dies,” does it make
you sad? For lots of people, the answer is “yes” — and that tells us something
important, and potentially worrisome, about our emotional responses to the
social machines that are starting to move into our lives.
For Christal White, a 42-year-old marketing and customer
service director in Bedford, Texas, that moment came several months ago with
the cute, friendly Jibo robot perched in her home office. After more than two
years in her house, the foot-tall humanoid and its inviting, round screen
“face” had started to grate on her. Sure, it danced and played fun word games
with her kids, but it also sometimes interrupted her during conference calls.
White and her husband Peter had already started talking
about moving Jibo into the empty guest bedroom upstairs. Then they heard about
the “death sentence” Jibo’s maker had levied on the product as its business
collapsed. News arrived via Jibo itself, which said its servers would be
shutting down, effectively lobotomizing it.
“My heart broke,” she said. “It was like an annoying dog
that you don’t really like because it’s your husband’s dog. But then you
realize you actually loved it all along.”
The Whites are far from the first to experience this
feeling. People took to social media this year to say teary goodbyes to the
Mars Opportunity rover when NASA lost contact with the 15-year-old robot. A few
years ago, scads of concerned commenters weighed in on a demonstration video
from robotics company Boston Dynamics in which employees kicked a dog-like
robot to prove its stability.
Smart robots like Jibo obviously aren’t alive, but that
doesn’t stop us from acting as though they are. Research has shown that people
have a tendency to project human traits onto robots, especially when they move
or act in even vaguely human-like ways.
Designers acknowledge that such traits can be powerful
tools for both connection and manipulation. That could be an especially acute
issue as robots move into our homes — particularly if, like so many other home
devices, they also turn into conduits for data collected on their owners.
“When we interact with another human, dog, or machine,
how we treat it is influenced by what kind of mind we think it has,” said
Jonathan Gratch, a professor at University of Southern California who studies
virtual human interactions. “When you feel something has emotion, it now merits
protection from harm.”
The way robots are designed can influence the tendency
people have to project narratives and feelings onto mechanical objects, said
Julie Carpenter, a researcher who studies people’s interaction with new
technologies. Especially if a robot has something resembling a face, its body
resembles those of humans or animals, or just seems self-directed, like a
Roomba robot vacuum.
“Even if you know a robot has very little autonomy, when
something moves in your space and it seems to have a sense of purpose, we
associate that with something having an inner awareness or goals,” she said.
Such design decisions are also practical, she said. Our
homes are built for humans and pets, so robots that look and move like humans
or pets will fit in more easily.
Some researchers, however, worry that designers are underestimating
the dangers associated with attachment to increasingly life-like robots.
Longtime AI researcher and MIT professor Sherry Turkle,
for instance, is concerned that design cues can trick us into thinking some
robots are expressing emotion back toward us. Some AI systems already present
as socially and emotionally aware, but those reactions are often scripted,
making the machine seem “smarter” than it actually is.
“The performance of empathy is not empathy,” she said.
“Simulated thinking might be thinking, but simulated feeling is never feeling.
Simulated love is never love.”
Designers at robotic startups insist that humanizing
elements are critical as robot use expands. “There is a need to appease the
public, to show that you are not disruptive to the public culture,” said Gadi
Amit, president of NewDealDesign in San Francisco.
His agency recently worked on designing a new delivery
robot for Postmates — a four-wheeled, bucket-shaped object with a cute, if
abstract, face; rounded edges; and lights that indicate which way it’s going to
turn.
It’ll take time for humans and robots to establish a
common language as they move throughout the world together, Amit said. But he
expects it to happen in the next few decades.
But what about robots that work with kids? In 2016,
Dallas-based startup RoboKind introduced a robot called Milo designed
specifically to help teach social behaviors to kids who have autism. The
mechanism, which resembles a young boy, is now in about 400 schools and has
worked with thousands of kids.
It’s meant to connect emotionally with kids at a certain
level, but RoboKind co-founder Richard Margolin says the company is sensitive
to the concern that kids could get too attached to the robot, which features
human-like speech and facial expressions.
So RoboKind suggests limits in its curriculum, both to
keep Milo interesting and to make sure kids are able to transfer those skills
to real life. Kids are only recommended to meet with Milo three to five times a
week for 30 minutes each time.
Comments
Post a Comment