3 Ways We Can Be Tricked Into Liking Robots

Research sheds light on how we experience artificially intelligent machines

Posted Jan 11, 2021

Kiosea39/Dreamstime
Source: Kiosea39/Dreamstime

Service robots interact with us in retail, hospitality, healthcare, and warehouse settings. They can come in "humanoid" forms and they can be fully or semi-autonomous. 
 
The market for service robots has been strong and is only growing. In fact, Covid-19 has contributed to the market's growth. Robots have helped with disinfection, manufacturing, and home delivery. 

Researchers have begun to explore the psychology behind our interactions with robots. 

One team recently published a study in the Journal of Applied Psychology. It looked at the hidden factors which affect the way we respond to service robots. The results are telling.

The researchers wanted to find out whether we would be more satisfied with service robots that are "anthropomorphized"—or, in other words, with robots that have human features.

To find out, they started with an AI-enabled robotic arm capable of learning and performing tasks such as painting and packing items and inserting printed circuit boards.

There were three features of the robot that the team manipulated in the study:

1. Voice: Either a female voice in an American accent in the anthropomorphism condition or a mechanistic voice in the control condition.
2. Name: Either "Allison" in the anthropomorphism condition or “Robotic Arm 57174” in the control condition.
3. "Face": Either a face displayed on the screen with moving lips while “speaking” in the anthropomorphism condition or a blank screen in the control condition. 

The researchers recruited undergraduate business students at a large university in Southeast Asia. When a participant entered the lab, the robot introduced itself and asked them to complete some initial survey questions.

When finished, the participant told the robot "I am done." At that point, the robot asked the participant to choose one of two snacks: potato chips or a chocolate bar. 

The experiment was set up so that the robot would sometimes give the participant the wrong snack. 

After the snack was delivered, a research assistant entered the lab and asked the participant to complete a customer satisfaction survey. They were asked to agree or disagree with the following:

  • “I feel satisfied with the robot" 
  • "I found real enjoyment during my interactions with the robot”  
  • “I  consider the robot to be rather pleasant” 

The researchers found that, in general, the robot received higher customer satisfaction ratings when it was anthropomorphized. 

This difference was driven by the perception of the robot as possessing enhanced experience and agency.

The team also found that, in the condition in which the robot was set up to give the wrong snack, participants who interacted with the anthropomorphized version of the robot were less dissatisfied than the participants who interacted with the version that was not anthropomorphized.

A future is creeping up on us in which our daily life will be replete with interactions with artificially intelligent robots.

What can we take away from this study? 

It all depends on what we are interested in. 

These researchers see their work as contributing to the design of future service robots that can deliver greater customer satisfaction. For this reason, they argue that service robots should be assigned names, a human voice, and a face.  

But we may also look at this from a different angle. We can ask: What are the implications for human happiness if we increasingly interact with robots that we are tricked into perceiving as human?

It is, of course, too soon to know the long-term consequences. 

But we may find that it becomes common for us to try to forge the kinds of connections with machines that we would with human beings.

The risk is that this will lead to our suffering from feelings of disconnection that we do not fully understand and that society will face an epidemic of loneliness even more grave than the one we face today.