"My focus is emotion-aware AI tutors for education; combining affective computing with large language models to personalize learning and advance equity." — Chenyu Zhang, Harvard University​

A New Frontier for Intelligent Learning

For decades, the promise of AI in education has centered on efficiency, offering faster feedback, personalized quizzes, and tireless 24/7 support. But in 2026, the narrative is shifting, thanks in no small part to Chenyu Zhang, whose work asks whether machines might not only teach but also help learners feel truly seen and heard. Zhang's work brings together large language models and affective computing to explore how AI can make learning more responsive, personalized, and equitable.

Growing up in a small town in China and later studying in Toronto and Cambridge, Zhang intimately understood the gaps that traditional classroom settings left behind—moments when emotional struggles went unnoticed, and frustration stymied potential. "Emotion is the silent driver of learning," Zhang reflects. "If teachers and AI systems ignore it, they leave students behind." These lessons propelled him toward a series of research positions at MIT Media Lab, Stanford HAI, and Harvard, where he led pioneering studies on emotion-aware tutoring.​ Zhang's work is shaped by a rare combination of technology and education. He studied Computer Science and Artificial Intelligence at the University of Toronto. He was also trained as an educator at Harvard University, allowing him to bridge advanced technical systems with the realities of teaching and learning. With a decade of coding experience and three years of teaching computer science, he combines strong engineering foundations with practical classroom insight.

The Science of Feeling: Affective Computing Meets AI Tutoring

Zhang's breakthroughs rest on marrying affective computing—the study of technology that can recognize and respond to human emotions—with the computing power of LLMs. In his widely cited ACII 2025 paper, Zhang introduced the ensemble-LLM framework, the first of its kind to capture, analyze, and respond to affective dynamics across numerous real tutoring conversations. The heart of this method is simple yet profound. Instead of relying on a single AI perspective, Zhang blended multiple leading models (Gemini, Claude, GPT) to triangulate emotional cues and refine feedback.​

Data from Zhang's studies shed light on the emotional landscape of student–AI interactions. Across 16,986 dialogue turns with 261 students in three U.S. institutions, "emotional inertia" emerged as a defining pattern. Positive moods tended to persist through 57% of consecutive tutoring turns, while negative emotions lingered almost half the time; a nuanced picture that challenged assumptions about AI's capacity to motivate. "Rapid rebounds and fragile emotional persistence demanded new tutor designs that could sense both confusion and momentum, adjusting approach in real time," Zhang writes.​

These insights didn't remain academic. Zhang's research influenced the development of GlowingStar, an AI-powered tutor now debuting in American classrooms, built on the very principles he established: real-time emotion analysis, transparent explanations, and feedback tailored not just to knowledge but to student well-being.​

Equity at Scale: Personalization and Global Impact

As AI tutoring sweeps across continents, the question of equity comes sharply into focus. Zhang's approach isn't content with uniform progress; it instead insists on adaptive support for every student, regardless of background. "The real power of personalized learning is not in customizing pace or difficulty, but in responding to the emotions that drive each student's experience," he explains.​

By 2030, emotion-aware AI tutoring is expected to underpin more than 59% of the digital education market, with North America, the Asia-Pacific region, and Africa leading the growth and adoption. Zhang's research highlights the significant impact of step-by-step scaffolding and multilingual support, which is crucial for communities that traditional systems have historically underserved.​

The numbers back up the vision: Global EdTech revenue from emotion-aware AI reached $3.7 billion in 2025 and is forecast to quadruple by the end of the decade, with standalone tutoring platforms driving early adoption. Notably, AI engines built on Zhang's ensemble protocols have demonstrated a 16% improvement in engagement and a 27% reduction in dropout compared to rule-based systems, all while documenting feedback decisions for educators to review, helping ensure inclusion, transparency, and trust at every turn.​

The Classroom in Practice

Zhang's influence extends well beyond the lab. As a teaching assistant at MIT, he has guided students in multimodal HCI, emotional intelligence, and affective computing, nurturing the practical skills and ethical curiosity necessary for tomorrow's AI engineers. At Harvard, he applies these lessons to teaching formative feedback protocols, utilizing generative AI and role-based agents to scale reflection, inclusion, and genuine personal growth.​

His teaching record spans hundreds of hours in virtual and physical classrooms, where he has codified AI curricula, designed prompt engineering workshops, and led peer mentorship programs from Toronto to Chengdu. Collaborators at Stanford and MIT credit him with catalyzing new discussions on responsible AI, highlighting the risks and rewards of emotion-driven systems. "Affective computing isn't just a technical skill," Zhang asserts. "It's a community design problem, asking us to rethink both technology and pedagogy so every learner feels supported."

Projects like Code in Place and AI4ALL demonstrate Zhang's commitment to scaling equity, teaching and mentoring over 2,000 graduate students, and guiding dozens of teams in rapid prototyping and launch of AI solutions. "Every learner is a glowing star," he says, echoing his startup's mission. "The goal is not to outscore, but to outshine—by nurturing curiosity and resilience across borders and cultures."

Criticism, Accountability, and the Meaning of Empathy

The ambition of emotion-aware AI tutors provokes both skepticism and admiration. Dr. Annette Bell, a psychologist and critic of AI in education, warns: "Simulated empathy in machines risks substituting algorithmic encouragement for genuine human care and understanding. The danger is that step-by-step advice and emotional analysis could be mistaken for mentorship, especially if systems don't make their artificial nature clear."

Zhang welcomes the critique but insists on principles of transparency and agency: "A good tutor—human or machine—invites questions and challenges its own feedback. We should design systems that openly show their reasoning, let students disagree, and frame reflection as a strength, not a flaw." His method embraces uncertainty and disagreement, allowing learners to see not only the answers but also the reasoning behind the models' feedback, cultivating resilience and independent thinking rather than dependency.​

As emotionally intelligent AI systems gain ground, experts, critics, and educators alike call for standards around consent, data ethics, and fairness. Zhang's protocols, already cited in policy discussions and educational conferences, set a high bar. He emphasizes auditable decisions and externally validated annotation, so equity isn't just promised, but measurable.​​

Future Horizons: The Shape of Learning

Zhang's vision for education is expansive yet grounded in practical wisdom. His work draws on collaborations with researchers across North America, robust scientific inquiry, and the lived experience of classroom silence and breakthrough moments. Publications from 2024 to 2025 describe the evolution of affective feedback models, multimodal diagnostic protocols, and adaptive curriculum design, each raising the bar for what emotion-aware AI can accomplish.​

Forecasts through 2030 suggest that AI tutors designed to be more human-centered, emotion-aware, and context-aware will not only reshape learning in schools but also influence workforce adaptation and lifelong skill development. Zhang argues that the next leap won't be technical alone, but social: "Participation is as important as performance. When reflection, resilience, and respect become operational standards, equity moves from aspiration to reality."​

Yet the future remains open-ended. As AI-powered platforms begin to "read the room," Zhang reminds educators and technologists to stay humble: "Emotion-aware AI will never replace what human teachers and peers bring: compassion, intuition, mentorship. But it can help widen circles, catch moments of doubt, and celebrate progress for every learner."

Reflections on a Changing Landscape

Zhang's career mirrors the evolution of AI in education, bridging cultures, disciplines, and emotional understanding. From small-town classrooms in China to the labs of Stanford, Harvard, and MIT, his work insists that the true test of intelligent machines lies not only in how they teach, but how they listen.

As emotion-aware AI tutors move closer to becoming classroom staples, Zhang's research continues to shape conversations about transparency, empathy, and equity. Through his efforts, the boundary between feeling and intelligence grows ever more fluid, inviting each student, in his words, "to be a glowing star."