New University of Minnesota report raises important questions about artificial intelligence in education, with findings that have particular relevance to an unpublished doctoral study involving 233 Guyanese learners.

GEORGETOWN, GUYANA | October 19, 2026

A major new international research report is challenging some of the assumptions driving the rapid introduction of artificial intelligence into classrooms, warning that while AI can strengthen tutoring, feedback and academic practice, excessive reliance on the technology may undermine the very thinking skills education is intended to develop.

The October 2026 expert consensus report, published by the University of Minnesota’s College of Education and Human Development, brought together 30 learning scientists and education researchers to examine how generative AI could affect 65 distinct learning processes. The experts reached consensus on 42 of those processes, identifying both opportunities for educational improvement and significant risks to student development.

For developing countries such as Guyana, where digital learning technologies are increasingly being considered as solutions to educational inequality, the findings raise an important question: Does the introduction of artificial intelligence actually improve student learning, or does it simply make educational tasks easier to complete?

According to the researchers, the distinction is critical.

Where AI Helps and Where It May Hurt

The report identifies what researchers describe as the GenAI Paradox. Artificial intelligence appears particularly useful during the practice and reinforcement stages of learning, when students have already been introduced to new concepts and need opportunities to strengthen their understanding.

AI-powered tutors can provide explanations, generate practice questions, offer immediate feedback and help students work through difficult material at their own pace.

However, the experts expressed concern about allowing AI to take over the early stages of knowledge acquisition or the more advanced stages of independent reasoning and problem-solving.

When students rely on AI to generate answers before attempting to understand a subject themselves, they may bypass essential mental processes involved in remembering, evaluating and applying information.

The report identified risks involving critical thinking, creativity, memory formation, self-regulated learning, motivation and collaboration. Younger learners were considered particularly vulnerable because many are still developing the foundational knowledge and independent learning habits needed to evaluate AI-generated information.

The experts also cautioned that AI interactions should not replace relationships with teachers and classmates, which remain essential to children’s social and emotional development.

Guyanese Doctoral Research Offers Important Local Evidence

The international findings are particularly relevant to research conducted in Guyana by Dr. Karen Abrams, Director of STEMGuyana and founder of Abrams & Associates Research.

In her unpublished 2026 doctoral dissertation at the University of Florida, titled Evaluating the Impact of AI-Driven Tutoring on Academic Outcomes and Engagement in STEMGuyana’s Learning Pods, Abrams investigated whether an AI-enhanced educational platform produced stronger academic outcomes than a comparable platform without AI features.

The eight-week study involved 233 learners participating in STEMGuyana’s community-based Learning Pods.

Students were assigned by Learning Pod to one of two learning environments. The AI-supported group, comprising 105 learners, used Pathway Online, which incorporated an AI tutor, automated quizzes and text-to-speech support. The comparison group, comprising 128 learners, used K12ed, a non-AI learning platform. Both groups received curriculum-based instruction and facilitator support.

Students completed Mathematics and English assessments before and after the intervention.

The findings showed that both groups improved academically, but the AI-enhanced group did not achieve statistically significant learning advantages over the non-AI group.

In Mathematics, the average gain was 12.52 points for the AI-supported learners compared with 12.82 points for the control group. In English, the AI group gained an average of 11.55 points compared with 9.93 points for the control group. Neither difference was statistically significant.

The study also found that measured use of individual AI features and overall platform time did not significantly predict academic improvements.

These findings do not establish that AI is ineffective or harmful. The study examined a relatively short intervention, and longer-term outcomes may differ. However, they challenge the assumption that adding AI capabilities to an educational platform will automatically produce better academic results.

The Guyanese study also provides a useful distinction from the international report. While the University of Minnesota experts focused largely on general-purpose generative AI systems, the STEMGuyana research examined AI features incorporated into a structured educational platform.

Together, the studies support the need to evaluate AI according to how it is used, the learning environment in which it operates and the measurable outcomes it produces.

Implications for Guyana’s Education System

The findings carry implications for Guyana as schools, educational organisations and technology developers explore ways to expand learning opportunities through artificial intelligence.

Digital education has particular appeal in communities facing shortages of specialist teachers, limited access to private lessons and geographical barriers to educational services.

AI-supported platforms could help students receive additional practice and explanations beyond the classroom. They could also assist teachers with identifying learning difficulties and preparing appropriate instructional materials.

However, technology adoption must be accompanied by evidence of effectiveness.

Abrams & Associates Research identifies several priorities arising from the international report and the Guyanese study.

Students should be encouraged to attempt problems independently before receiving AI assistance. AI tutors should guide learners through explanations and questions rather than routinely supplying complete answers. Teachers and facilitators should remain involved in monitoring understanding, motivation and academic progress.

Educational assessments should also distinguish between work completed with AI assistance and knowledge students can demonstrate independently.

The findings are especially relevant for primary-school learners, who require sustained opportunities to develop reading, writing, numeracy and reasoning skills before becoming heavily dependent on automated assistance.

For educational policymakers, the research suggests that investments in AI should be evaluated against existing educational approaches, including teacher-supported instruction and well-designed digital learning systems without generative AI.

A platform’s technological sophistication, number of features or level of student engagement cannot, by itself, establish its educational effectiveness.

Experts Call for Stronger Research

One of the University of Minnesota report’s most consequential findings concerns the quality of existing research on AI in education.

Despite growing investment and widespread adoption, the researchers concluded that much of the available evidence remains insufficiently rigorous. Many studies involve small samples, short observation periods or comparisons that fail to isolate the actual contribution of AI.

The report calls for longer-term studies, stronger research designs and more careful examination of how AI affects different learners and learning processes.

This is particularly important for developing countries, where educational technologies are sometimes adopted with limited independent evaluation of their effectiveness in local classrooms.

The unpublished STEMGuyana research contributes evidence from a Guyanese, community-based learning environment, while also illustrating why additional studies are necessary to understand which AI features are effective, for which learners and under what conditions.

The emerging message from both bodies of research is that the future of AI in education should be determined by improvements in learning, rather than the sophistication of the technology itself.

Artificial intelligence offers genuine opportunities to expand academic support, personalise practice and assist teachers. Its educational value will ultimately depend on whether students are becoming more knowledgeable, more capable of solving problems and more independent in their thinking.

For Guyana, that is the standard against which AI investments in education should be measured.

About Abrams & Associates Research

Abrams & Associates Research is a Guyana-focused research and consultancy initiative led by Dr. Karen Abrams, whose work spans educational technology, artificial intelligence, academic achievement, digital transformation and evidence-based programme evaluation.

Research referenced: Kendeou, P., Greene, J. A., Nixon, N., et al. (2026). Ways Generative AI Can Support and Threaten Learning in K-20 US Education. Expert Consensus Report, University of Minnesota College of Education and Human Development.

Abrams, K. (2026). Evaluating the Impact of AI-Driven Tutoring on Academic Outcomes and Engagement in STEMGuyana’s Learning Pods. Unpublished doctoral dissertation, University of Florida.