Purdue researchers are advancing robotics by teaching AI systems to understand and respond to human emotions and nonverbal cues, paving the way for more empathetic, human-like interactions. (Source: Image by RR)

Scientists Aim to Make Robots More Like Companions and Less Like Tools

A Purdue University research team is working to make robots more socially intelligent by improving their ability to understand and respond to human emotions, tone, and nonverbal cues. Led by computer scientist Sooyeon Jeong, the effort focuses on bridging the gap between how machines communicate and how humans naturally interact—moving beyond simple question-and-answer exchanges toward more fluid, human-like conversations.

The research, as noted in an article at purdue.edu, centers on “active listening,” a key element of human communication that includes subtle behaviors like nodding, tone matching, and verbal cues such as “right” or “I see.” These signals—known as backchannels—help people feel heard and understood. Jeong’s team is training robots using large language models and real human conversation data to replicate these behaviors, allowing machines to respond more naturally and empathetically in real-time interactions.

One practical application is a robot “study buddy” designed to help users stay focused and productive. In experiments, robots offered varying levels of support—from simply working alongside students to providing reminders and emotional encouragement. Results showed that effectiveness depends heavily on individual personality, mood, and context, suggesting future robots will need adaptive social intelligence that can read situations and adjust behavior dynamically.

Beyond education, the research has implications for healthcare, therapy, and elder care, where emotional comfort and trust are critical. By making robots more like attentive companions—and less like impersonal tools—Jeong’s work aims to ensure that as AI becomes more embedded in daily life, it enhances human well-being through meaningful, supportive interactions.

read more at purdue.edu