In much the same way that a human gains knowledge and progresses as it grows, Wanli Xing views an artificial intelligence agent as a “learning buddy” that matures and heightens its understanding through continual interaction with students.
“From an emotional or knowledge perspective, initially, AI doesn’t know a lot. But it can learn with you, grow with you, and work with you to solve problems,” said Xing, who joined the University of Miami in January with a double appointment as a professor with the School of Education and Human Development and with the Frost Institute for Data Science and Computing.
Xing’s research, which is extensively involved in AI in education, will strive to empower educators across the University to harness classroom data in new ways that improve teaching and deepen student engagement.
His perspective of exploring AI as a learning buddy is particularly relevant to his research of student-tutor interactions, and he credited the emergence of large language models (LLMs), an aspect of the latest phase of AI development, with revolutionizing the educational landscape in terms of this relationship.
“Earlier generations of educational AI were largely based on rule-driven and traditional machine learning-based systems, which limited how intelligently an AI tutor could interact with students. You could easily see the difference between an AI tutor and a human one,” he noted. “With the emergence of large language models such as ChatGPT and the development of AI agents, we are now seeing a fundamental shift in how AI can support learning and interact with students.”
His current research project is analyzing discussion patterns and tutoring strategies that better engage students in mass or STEM areas of learning, often challenging for students.
“The AI-powered tutor becomes another teacher to help you. Then we flip or reverse the role of AI,” Xing explained. “AI needs help to become a buddy of yours, to become a junior of your needs, and to tap into your engagement like a conscience or your human nature try to help somebody. We ask AI for help, and AI asks us for help.
“Students actually improve their engagement and their understanding of mathematical concepts through learning by teaching, traditionally a very successful strategy for students. They must learn and master concepts in order to teach them [to AI buddy],” he said.
The enhanced tutor-student relationship is just one aspect of the advances AI has brought to the field of education, he emphasized.
Xing described his overall research in three areas. First, creating AI technologies to support education, design development, and testing tools for K-12 classrooms and also higher education.
Second, data mining and learning analytics research of the mountains of big data that has accumulated on the multitude of education platforms. “I’m trying to understand the fundamental processes for learning to take place: What kind of learning behaviors and processes can lead to better engagement and learning outcomes?” Xing explained.
The third realm is more interdisciplinary, focused on future workforce development, particularly advanced technology areas. He’s studying how to teach undergraduate students to better understand AI and data science generally, as well as more advanced topic areas such as quantum computing, along with mastering the skills of semiconductors.
Xing emphasized the vast amount of AI research that has taken place over the past 20-30 years and pointed to professional societies such as the International Artificial Intelligence in Education Society, the Society for Learning Analytics Research, the International Educational Data Mining Society, the IEEE, the global community for technology, and others that have contributed to advances.
Xing’s research focuses on students’ technology use, learning impacts, and emerging AI-driven tools. He admittedly has little experience with social media, but he expressed his concern for Australia’s legislation enacted last December that banned social media for kids under 16.
He urged more regulations for content and a more moderate approach while highlighting that media literacy is part of the future competencies for teens, middle school, and secondary students when they get to college or enter society.
“One thing that we have to admit is that we are living in a digital world. We are the future generation, whether we like it or not, and we will have more and more demand on digital and media literacy skills, now referred to as ‘AI literacy,’” said Xing, who holds a Ph.D. in information science and learning technologies from the University of Missouri.
Xing suggested the creation of a “United Nations for AI.”
“AI is controlled by a few big tech companies who make all the decisions on how AI behaves and how AI impacts society. But there are lots of transparency, ethical, and regulatory issues,” he noted. “Right now, we’re still in the ‘explosive’ phase of the technology—we don’t have much regulation on how AI should work or interact or be present in future society.”
He noted that there is generally a lack of regulatory clarity for every new technology in its first few years.
“We just need to think more, continue to work and research, and join the journey,” Xing said. “Whenever there’s a revolution, we can’t reject what’s coming.”
