© IE Online Media Services Pvt Ltd
Tags:
Journalism of Courage
Scientists have found that robots could prove useful as mental wellbeing coaches for employees in offices. But the employees’ perceptions of how effective the robots are depends a lot on how they look.
Researchers from the University of Cambridge conducted an “in-the-wild” study in a tech consultancy firm where they used two different robots as wellbeing coaches. As many as 26 employees at the firm participated in the weekly robot-led wellbeing sessions for four weeks. Interestingly, the participants who worked with a toy-like robot felt more of a connection with their coach than those who worked with a humanoid robot.
Both the robots had the exact same voices (Amazon’s AWS Polly’s Amy voice), similar facial expressions and scripts for these sessions. But the vastly different physical appearances of the two robots changed the way in which participants interacted with it. The results of the research will be presented today at the ACM/IEEE International Conference on Human-Robot Interaction in Stockholm.
To test the efficacy of the robot mental wellbeing coach, the researchers worked with Cambridge Consultants, a technology firm in the same town as the university. During the four week program, the employees were taken through four different wellbeing exercises by one of two robots: the QTRobot (QT) or Misty II robot.
QT is a humanoid robot that is almost childlike and is around 90 centimetres tall. Misty is a toy-like robot that measures 36 centimetres. Both the robots had screen faces that could be programmed with different facial expressions.
The researchers worked with different mental well-being coaches to program the robots with the same scripts and voices, and similar facial expressions and gestures during the sessions. The participants were then taken through different positive psychology exercises by a robot in an office meeting room.
Each of these sessions began with the robot asking participants to recall a positive experience or to describe something in their life they were grateful for. After this, the robot would ask them follow-up questions. After the sessions, the participants were asked to assess the robot with a questionnaire and an interview.
Not only did the participants who worked with the toy-like Misty robot report that they had a better working connection than those who worked with the child-like QT robot, but they also had a more positive perception of Misty overall. According to the researchers, this could be because Misty was able to match their expectations from a robot because it was more toy-like.
In the paper, the researchers wrote that the employees’ expectations of robot capabilities are heavily distorted by how they are portrayed in videos. The employees said that they expected more from these robots because of “demos from cutting edge teams” and that Alexa and Google Assistant drove their expectations.
According to the researchers, the most common response they got from participants is that the latter’s expectations of the robot did not match with reality. The researchers programmed the robots with a script but the participants hoped that there would be more interactivity. But it is quite difficult to create a robot that is capable of natural conversation. New developments in large language models, as seen with ChatGPT and the upcoming GPT-4, could really be helpful here.
Also, earlier research has already established that perceptions of a robot’s behaviour are heavily dependent on people’s demographic background.
“We agree that in the future, robots should be able to adapt to various contexts, also within different demographics. For example, different age groups may have different levels of knowledge of technologies, and different geographical locations may have different social norms,” explained Minja Axellson, co-author of the Cambridge research, to indianexpress.com in an email.
The eventual goal of the researchers is to create a fully-autonomous mental well-being coach that can be used in various contexts in the real world. The goal is not to replace an actual human coach, but rather give the option to organisations that do not have the resources to hire one for their employees.
“Future research could, for example, examine how different social norms should be accounted for in the robot’s behaviour during coaching, and if the robot should express more or less detailed instructions to interact with it for different age groups,” said Axellson.
Even if the robot is not as effective as an actual coach, it can help serve as a physical reminder to practice wellbeing exercise. And according to the researchers, even just saying things out loud to the robot can be helpful when trying to improve mental wellbeing.
Now, the team is working on enhancing the robot coaches’ responsiveness during the coaching practices and interactions.
