Yuan Sun and Shyam Sundar of Penn State have published a study demonstrating an “IKEA effect” when you set up your own robot: In their careful study, putting the robot together yourself rather than watching it put together led to a “sense of ownership” and a “sense of accomplishment”, as well as a recognition of the “cost” of doing the work. In addition, this led to a higher evaluation of the robot itself, especially if it was framed as oriented toward a task rather than interaction with the user.
In one sense, this seems like an obvious finding, given that there are plenty of cases, including DIY flat pack furniture, maker spaces, home gardening, and every DYI project in history. We like things when we make them or participate in making them. Why would robots be different?
But this study is important for robotics because there is a tremendous amount of research being directed at designing robots that get along well with people. These efforts range from heroic efforts to build in social behaviors and ethics, to investigations of appearance and flight paths designed to be non-threatening. Many of these studies place the responsibility (and intelligence) in the robot, assuming that the people are passive, disinterested, and unteachable.
Sun and Sundar’s study finds that people evaluate robots and interactions with robots the same way they do everything else. And people actively learn and develop interest and positive feelings about robots. The robot doesn’t have to do all the adapting, if you design the human-robot system in the right ways.
Their study suggests that, no matter how the robot is designed, there is a big difference if the user feels attachment and investment in the particular robot they deal with. My own intuition is that this will also carry over to reducing feelings of threat and discomfort.
For example, I expect that a drone that I snapped together and launched will be less threatening than the exact same drone that flies the same paths near me, but isn’t “my” drone.
I would suggest that this effect can be enhanced further by giving the robot a name, and customizing it a little bit so it is easy to distinguish.
These things are really cheap, and require no heroic AI, and I bet they will work well. As Evan Ackerman comments, “even if you can sell your robot as fully assembled and ready to go right out of the box, people will like your robot better (and think that it’s a better robot) if you let them participate in the setup process”.
In cases where the robots are already assembled, I would suggest that users be allowed to pick their own robot out of the “orphan robot shelter”. (I would make each robot a little different, and perhaps scruffy looking, a la orphan puppies.) The user will pick his or her robot, give it a name, touch it, maybe “feed it” power, and launch it.
I think that this bit of theater would go far to not only reducing discomfort, but creating strong positive bonds—with very specific robots. Furthermore, the user will be able to tell “his” robot, which should be trusted, from “stray” robots which should be scrutinized carefully.
Again, this kind of interaction is really cheap, and doesn’t require extraordinary intelligence in the robot. After all, I love my own cat, even if she is kind of dumb or troublesome sometimes!
- Sun, Yuan and S. Shyam Sundar, Psychological Importance of Human Agency: How self-assembly affects user experience of robots, in The Eleventh ACM/IEEE International Conference on Human Robot Interaction. 2016, IEEE Press: Christchurch, New Zealand. p. 189-196.