-- Cambridge, MA — September 2025. GlowingStar today announced the development of the world’s first affective AI tutor, an emotion-aware learning coach designed to adapt to both cognition and emotion while making its reasoning transparent to educators. The launch positions GlowingStar founder Chenyu Zhang, a recent Harvard graduate, to compete directly with OpenAI’s newly released study mode.
OpenAI’s move underscores the industry’s shift toward step-by-step guidance and stronger pedagogy, but what still gets overlooked is how a learner actually feels in the moment. That’s the space GlowingStar is claiming. Its AI tutor infers signals such as curiosity, confusion, and frustration from student interactions and adapts in real time. If it senses struggle, it slows down and reframes. If it detects momentum, it raises the challenge. Just as important, the system’s reasoning is auditable—educators can see why it thinks a student is stuck and how it plans to respond.
Zhang’s conviction comes from classrooms he knows across China, Canada, and the United States. Growing up in a small city in Shandong and later studying and working in North America, he saw a recurring gap: emotions weren’t taken seriously in school, and teachers lacked the bandwidth to catch quiet frustration before it derailed progress. As a first-generation university student and immigrant, he felt that silence firsthand—and spent three years tutoring computer science, witnessing how a well-timed nudge could turn “stuck” into “I’ve got this.”
While completing his master’s at the Harvard Graduate School of Education, Zhang began conducting research across MIT Media Lab, Stanford HAI, the Harvard Berkman Klein Center, and Harvard Business School, focusing on affective computing, large language models, and AI-powered tutoring systems. To lay GlowingStar’s scientific foundation, he personally led a large-scale study of 16,986 tutoring dialogue turns across 261 students at three U.S. institutions. His ensemble-LLM method revealed how emotions like curiosity, confusion, and frustration ebb, flow, and rebound during problem-solving. This work, published at ACII 2025, now underpins GlowingStar’s product vision.
For Zhang, GlowingStar is more than a company—it’s a mission shaped by lived experience. He co-founded a Second-Year Learning Community at the University of Toronto, taught Python globally through Stanford’s Code in Place, mentored underrepresented students via AI4ALL, and co-organized the Decentralized AI Summit at MIT with over 1,000 participants. Each step reinforced the same insight: learners thrive when they feel seen, not just scored.
“Every learner is a glowing star,” Zhang says. “Our job is to help them shine—by noticing when they’re stuck, encouraging them when they’re close, and challenging them when they’re ready.”
As OpenAI pushes study-mode tutoring forward, GlowingStar is charting the next frontier: AI that can read the room—pairing step-by-step guidance with emotion-aware adaptation and explanations educators can trust.
About GlowingStar
GlowingStar builds AI tutors that combine large language models, affective computing, and human-centered design to deliver emotionally aware learning at scale. Founded by Harvard graduate Chenyu Zhang (chenyuzhang.com), with research experience at MIT Media Lab, Stanford HAI, and the Harvard Berkman Klein Center. Learn more at glowingstar.ai and chenyuzhang.com.
Contact Info:
Name: Chenyu Zhang
Email: Send Email
Organization: Glowingstar
Website: https://glowingstar.ai/
Release ID: 89170632
Google
RSS