What role does human-robot interaction play in the future of AI?

As artificial intelligence (AI) rapidly becomes more sophisticated, it is still common to presume that robots will not take the form of people, physically behave like humans, or coexist in the world as peers. This may be a good thing, as we will be able to tell the difference between people and robots in the real world even as things like deepfakes become harder to spot online. Yet researchers at Stevens are working to close the behavioral gap between robots and people. Electrical and Computer Engineering Professor Yi Guo is leading the research team to develop robots that mimic the behavior of humans in crowds and can blend in with groups of people.

Currently, most robots move awkwardly within crowds and around people based on programming that does not take into account people’s perceptions of rudeness. They sense all objects, living or nonliving, the same way. This can cause them to interrupt the flow of walkers, get too close to people, or just bring a general sense of discomfort to those around them.

To teach robots how to display appropriate social skills, Professor Guo “identifies patterns of pedestrian navigation behaviors and extracts social affinity features that map how individuals in a crowd move in relation to one another.” From there, she utilizes an “imitation learning algorithm” to program the robots to move in a respectful manner within crowds.

Furthermore, Guo’s team programmed their robots to have the ability to pick a person to label as a “leader,” and follow them at times when crowds are large and dense to keep up with the natural flow of the crowd; other robots will not know what to do in that kind of situation and will usually just shut down.

Professor Guo and her research team’s robots have the power to significantly benefit communities and the economy through help with emergency situations, crowd control, or even to assist in different sectors of the workforce.

The results of the team’s study are projected to be released in 2022, and through their work, Guo hopes to try and improve the relationship humans have with robots. She wants people to feel more comfortable and safe around robots and suggests that society can promote the acceptance of robots by bringing more publicity to robot development and learning more about how robots can help in many areas of life.

Professor Guo and her team’s work brings attention to an important question – what role does human-robot interaction play in the future of AI? Should projects such as hers continue and become more widespread, it is not farfetched to assume that robots will have the ability to seamlessly blend into crowds in the near future. The more exposure robots have to human actions, the closer they will be able to emulate our behavior. With this logic, and with enough development, it is very possible that AI will allow robots to act and speak according to the most common reactions people have in a situation.

Caution must be taken when exploring robotics and attempting to improve the abilities of AI, as the malleability of technology can lead to the development of harmful behavior. A notable example of this was seen nearly five years ago on Twitter when a Microsoft chatbot by the name of “Tay” was publicized with the ability to communicate with people. Via Twitter, Tay responded based on the typical replies of users. Trolls quickly taught the bot to frequently publicize very offensive comments, including political and racist statements. This was an alarming but predictable result of what happens when those with bad intentions have the ability to manipulate impactful technology.

As we continue to pursue the development of AI and the ability of robots to assimilate within society, it is important to make sure that robots are not misled to behave harmfully, and that the potential of negative instances such as Tay’s to dent the public perception of robots is not underestimated. The future power and influence of AI are up to how humans choose to interact with them, so we should try and promote discussion regarding robots and AI so that they are developed to be used in ways that are beneficial, not derogative, to society.