A primitive form of empathy. This is what a team of scientists from Columbia University (USA) have achieved thanks to a technology that could be described as the first brushstroke of empathy from a robot towards another robot.
The study, carried out in the Creative Machines Laboratory of Columbia Engineering led by Professor of Mechanical Engineering Hod Lipson, is encompassed in a broader objective to equip robots with the ability to understand and anticipate the objectives of other robots, only to starting from visual observations.
The researchers built a robot and placed it in an environment approximately 1 meter x 0.60 cm. They programmed the robot to search and move towards any green circle it could see. But there was a catch : sometimes, the robot could see a green circle in its camera and move directly towards it. But other times, the green circle would appear hidden by a tall red cardboard box, in which case the robot would either move to a different green circle, or it would not move.
Meanwhile, a fellow robot observed the scene.
After watching its partner spinning for two hours, the observation robot began to anticipate the target and the path of its partner, being able to predict the target and the path of its robot partner 98 times out of 100, in various situations, without explicitly told about the robotic companion’s visibility disadvantage.
“Our initial results are very exciting,” says Boyuan Chen, lead author of the study, which was carried out in collaboration with Carl Vondrick, assistant professor of computer science, and published in the journal Nature Scientific Reports. “Our findings begin to demonstrate how robots can see the world from another robot’s perspective. The observer’s ability to put himself in his partner’s shoes, so to speak, and understand, without being guided, whether or not his partner could see the green circle from his vantage point, is perhaps a primitive form of empathy. “
Like a couple of many years who can predict each other’s every move, the robot learned to predict the future actions and goals of its partner robot based on a few initial video frames. Certainly our ability to anticipate the actions of others makes it easier for us to live and work together successfully. Now, this type of communication has taken its first steps as far as robots are concerned, although, logically, the behavior exhibited by the robot in this study is much simpler than in human beings, but it may be the beginning of equipping robots with what they call “Theory of Mind,” a key hallmark of human and primate cognition, and a factor that is essential for complex and adaptive social interactions such as cooperation, competition, empathy, and deception.
Ethical doubts
Technology will make robots more resilient and useful, but when robots can anticipate how humans think, they can also learn to manipulate those thoughts.
“We recognize that robots are not going to remain passive instruction-following machines for long,” says Lipson, a co-author of the paper. “Like other forms of advanced AI , we hope that policy makers can help keep this type of technology in check, so that we can all benefit.”
Referencia:Visual Behavior Modelling for Robotic Theory of Mind. Boyuan Chen, Carl Vondrick and Hod Lipson. Scientific Reports (2021).DOI:https://doi.org/10.1038/s41598-020-77918-x
