Do humans see life in robots?
Are robots agents? Can robots be sentient?
Can robots show life?
One of the interesting debates in contemporary times is whether robots can ever be perceived as agents, having the independence to take actions and make sound moral judgement and decisions. One of the early positions on this is taken by Professor Selmer Bringsjord who in his paper titled “Ethical Robots: The Future Can Heed Us” argues that we don’t need to fear robots, we need to fear what people would do with robots because ultimately robots will do what humans program them to do. A counterpoint is offered by Professor John Sullins, who has argued that under certain circumstances, robots should be seen as moral agents, because they could potentially have autonomous thinking, and judgement to decide whether a decision is good or bad. Although Prof. Sullins views could have been labelled as science fiction when his paper was published in 2006, advances in Machine Learning (ML) have shown that it is possible for robots to learn things independent of what they are programmed to do, and also be able to judge and differentiate whether their action would be construed as good or bad. An interesting analysis of these two points of view can be seen in Florence Simon’s draft available here.
Can Cozmo show life?
In this context, there is some exciting new research by Hannah Pelikan, who researches on how robot sounds influence the interaction between humans and robots. In her latest work titled, “When a Robot comes to life”, Hannah looks at the various forms of agency (ability to act independently) that Anki Cozmo takes, as it interacts with a family of 3 over a video captured for 60 seconds. Cozmo’s interactions with the family are shown to be falling into the following different patterns:
Fully autonomous agency: In the greeting sequence, Cozmo’s is truly seen as a human counterpart, and the human greets Cozmo exactly in the same way as he would greet a human.
Hybrid agency: In the video, a brother uses Cozmo to play a prank on his sister by making Cozmo call his sister with a distorted name, and then making fun of it. While in this activity, Cozmo takes human form, it does so under instructions of another human who knows Cozmo’s abilities and skills, and can make due use of them.
Ascribed agency: At times, a human may think of Cozmo to have life, even though Cozmo is sitting idle and not doing anything. As an example, in the video, the sister is expecting a reaction from Cozmo such as a fist bump, even though she knows that Cozmo is in control of her brother.
Potential agency and non agency: At times, a human might want to test Cozmo and then come to the conclusion that Cozmo is a material object, and doesn’t have life.
Hannah’s work shows how Cozmo could transient across these forms of agency in a short span of 60 seconds. Hannah’s work argues that a human’s social responses to robots depends on their goodwill to use the robot in the specific situation and circumstance.
What do we think?
Hannah’s work is an interesting read, and shows how powerful the Cozmo and Vector robots really are, and how futuristic is the technology that was built into these robots. It’s great to see the use of such robots in research on Human Robot Interaction.