AI in Education: Striking Contrasts Emerge from New Research

The integration of artificial intelligence into education continues to be a topic of significant discussion, with new research constantly shaping our understanding. Recent findings from the University of Maryland, led by Jing Liu, director of the Center for Educational Data Science and Innovation, highlight the complex and sometimes contradictory effects AI tools can have on student learning and engagement. Two studies, released in October 2026, offer a compelling contrast, suggesting that the success of AI in education hinges not just on the technology itself, but on its design, implementation, and how students ultimately interact with it.

According to The Hechinger Report, Liu intended for these studies to be released together to underscore that AI in education is neither unilaterally positive nor negative. The field is actively working to understand how to best design tools that genuinely enhance learning over time, emphasizing the need to study student responses—including those unintended by creators.

Middle Schoolers Engage, Potentially Benefit from AI Math Tutor

One of the studies examined the use of an AI-powered math practice website called Goblins, implemented with 6,000 middle schoolers in Deer Valley, Arizona, during the 2025-26 school year. The platform allowed teachers to assign problems, and an AI tutor provided step-by-step assistance through text chat, a virtual whiteboard, and even spoken conversation. Teachers were encouraged to use it with struggling students, though some integrated it across their entire classes.

The Hechinger Report notes that students who completed at least two problems per week—totaling over 50 during the school year—showed significantly higher scores on year-end standardized math assessments compared to similar students who used Goblins less frequently or not at all. Accelerate, an independent nonprofit research organization that commissioned the study, estimated this statistical difference to be equivalent to approximately three additional months of learning. While the study cannot definitively prove Goblins caused these gains, as highly motivated students might have performed better regardless, the engagement rate itself is noteworthy.

A significant finding, according to The Hechinger Report, is that half of the 6,000 students met the usage threshold associated with higher math scores. This contrasts with the common “five-percent problem” in education technology, where often only a small fraction of students consistently use tools as intended. Liu suspects that Goblins' ability to allow students to converse aloud with a cartoon avatar of their teacher, and its multilingual support (including Spanish), contributed to this higher engagement. Teachers' control over assignments also appeared to be a factor, allowing them to tailor problems and even upload their own, echoing findings from earlier education technology research.

For parents and teachers considering AI tools for K-12 math practice, this study offers encouraging insights into designs that foster engagement. Tools that offer personalized, step-by-step guidance and adapt to student needs can be particularly valuable. Free resources like COSMIQ, a voice-driven AI tutor, offer similar interactive support for K-12 students across various subjects, providing accessible learning assistance.

University AI Assistant Linked to Lower Grades and Reduced Engagement

In stark contrast, Liu’s other experiment at the University of Maryland explored the impact of a university-developed AI assistant on college students. Prior to a potential university-wide rollout, Provost Jennifer King Rice commissioned Liu to study its effects. During the fall 2025 semester, 15 professors integrated the AI assistant into their courses, while another 15 taught similar courses without it. The subjects ranged widely from English and math to biology and social sciences.

The AI chatbot, pre-loaded with syllabuses, readings, and assignments, was available 24/7 via Canvas, the university’s online learning management system. Students could use it to ask about assignment instructions, deadlines, or to gain a better understanding of concepts. The Hechinger Report indicates that by the end of the semester, the 1,100 students in AI-enhanced classes had, on average, lower grades—a reduction of 0.11 points on a four-point scale. These students also interacted less with course materials and were less likely to attend office hours or meet with instructors and peers. The negative effects were even more pronounced for first-generation students.

Only 15 percent of students in the AI-enhanced classes ever used the assistant, and those who did primarily sought answers rather than utilizing its tutoring features. Liu and Rice, according to The Hechinger Report, hypothesize that the AI assistant’s constant presence might have inadvertently signaled that AI use was encouraged, potentially leading students to use external AI tools like ChatGPT to complete assignments, thereby learning less themselves. Provost Rice expressed particular concern that "AI tools could exacerbate inequities in learning outcomes."

Key Takeaways for Educators and Families

These contrasting findings underscore the critical importance of thoughtful design and implementation when integrating AI into education. For K-12 educators and parents, the Goblins study suggests that AI tools designed for interactive, personalized practice, especially those that allow for natural language interaction and teacher oversight, can be beneficial. The ability to speak aloud to an AI tutor, as offered by platforms like COSMIQ, can make learning feel more accessible and engaging, particularly for students who might struggle with traditional methods or English language learners.

The university study, however, serves as a cautionary tale. It highlights the potential for unintended consequences, even with well-intentioned tools. Simply making an AI assistant available doesn't guarantee positive outcomes; in some cases, it might inadvertently deter deeper engagement and critical thinking. Educators, parents, and students should be mindful of how AI tools are presented and used. Encouraging active learning, critical thinking, and direct engagement with course material remains paramount, regardless of the technological support available.

The research from the University of Maryland provides valuable insights into the evolving landscape of AI in education. It reinforces that effective AI integration requires careful consideration of student psychology, learning objectives, and potential behavioral shifts, ensuring that technology truly serves to enhance, rather than hinder, the learning process.