As artificial intelligence (AI) continues to reshape industries worldwide, higher education—particularly in science, technology, engineering, and mathematics (STEM)—stands at a crucial turning point. Leading this transformation is the AI4STEM Education Centre at the University of Georgia (UGA), where researchers and educators are harnessing AI to elevate teaching, learning, and assessment in science education.
At the core of this initiative is an ambitious goal: to create tools that help teachers engage students in authentic scientific practices while easing their workload. “Teachers are under immense pressure to align their instruction and assessments with next-generation science standards,” says Xiaoming Zhai, PhD, associate professor in the Department of Mathematics, Science, and Social Studies Education and affiliated assistant professor at UGA’s Institute for Artificial Intelligence. “Without time-saving tools—such as automated feedback or AI-supported scoring—teachers are less likely to implement high-quality assessments, and students ultimately miss out.”
From Scoring to Support: AI as a Teaching Partner
Much of Zhai’s early work at the centre revolved around automating assessment. His team developed machine learning systems capable of evaluating students’ scientific reasoning—whether in written explanations, data modelling, or other performance tasks—at a level of nuance once reserved for expert human scorers. These tools provided teachers with immediate, meaningful insights into how students were engaging with essential scientific practices.
But automation proved to be only the first step.
Today, Zhai is advancing the field further with “GenAgents”—AI-driven multi-agent systems designed to interact with students and educators in real time. These agents are being trained to support students through complex scientific tasks such as constructing arguments, analysing data, and developing explanations.
Imagine two students debating the cause of a temperature change in an experiment. One attributes it to a chemical reaction; the other believes it was a measurement error. A GenAgent could intervene not merely to provide the correct explanation, but to act as a facilitator, challenger, or moderator—depending on the situation.
“For high-performing students, the agent might encourage deeper reasoning,” Zhai explains. “For students who are drifting off task, it can gently steer them back to the scientific content.”
Expanding Access for Underrepresented Students
Another major project at the centre is siSTEMas, funded by the National Institutes of Health. This initiative uses game-based learning to introduce middle school students—especially those from underrepresented communities—to STEM concepts.
The project’s bilingual design sets it apart. By developing both English and Spanish versions of their science learning games, the team aims to ensure accessibility for multilingual learners. “We often see performance gaps rooted more in language proficiency than in science understanding,” Zhai notes. “By reducing language barriers, we hope to create equitable, high-quality learning experiences for all students.”
Challenges to Wider Adoption
Despite significant progress, integrating AI more broadly into STEM classrooms faces several hurdles.
First, public perceptions—often shaped by sensationalised media coverage—can influence how comfortable teachers and parents feel with AI tools. “People tend to hear more about the risks of AI than the benefits,” Zhai says. “That creates hesitation.”
Second, educators require professional development to confidently and ethically integrate AI tools into their instruction. Zhai’s team is developing a framework to guide this transition, helping teachers evolve from passive users of AI to active co-designers of AI-enhanced lessons.
Finally, interdisciplinary collaboration remains challenging. Bringing together experts from different fields requires not only infrastructure and administrative support but also a cultural shift toward valuing diverse contributions across academic disciplines.
Looking Forward
As the AI4STEM Education Centre advances its work, its mission remains rooted in a fundamentally human aspiration: empowering both educators and students to succeed in an increasingly complex world.
“I’m a teacher at heart,” Zhai reflects. “Preparing the next generation—not just with content knowledge, but with the confidence and creativity to think critically—that’s the most important work we can do.”
With innovations such as GenAgent, siSTEMas, and AI-powered assessments rapidly emerging, the future of STEM education promises to be more interactive, more inclusive, and more intelligent than ever before.
