Meet Henry Xie, student making AI more empathetic

Henry Xie, a 17-year-old student from Portland, is making waves in the world of artificial intelligence by developing a method that could make smaller AI models more empathetic. His innovation stands out because it doesn't rely on the massive computing resources needed by the biggest AI systems.

Henry, a student at Westview High School and member of the varsity swim team, created this project as part of the Regeneron Science Talent Search, a prestigious competition for high school students focusing on science and mathematics. He tackled the question of whether the empathy displayed by large AI models could be transferred to smaller, faster models.

His two-step approach involves first training smaller language models (SLMs) using examples of empathetic responses generated by large language models (LLMs) such as Gemini and ChatGPT. These responses were guided by psychological theory. Then, targeted prompts were used to differentiate stronger empathetic responses from weaker ones.

In trials, the trained SLMs were found to be more empathetic in at least 90% of cases compared to untrained SLMs. This success suggests that Henry's method could make empathetic AI more widely accessible, particularly as smaller models are often faster, cheaper, and more suitable for everyday applications where resources may be limited.

Beyond his AI project, Henry is also deeply involved in cybersecurity. He has been a three-time semifinalist in the CyberPatriot National Youth Cyber Defense Competition, one of the world's largest youth cybersecurity competitions. This expertise adds another dimension to his impressive technological background.