A new study from the University of Massachusetts has found that AI-generated lesson plans are not enhancing student engagement or developing critical thinking skills as expected. Despite the rapid integration of AI tools like ChatGPT, Gemini, and Copilot into classrooms, researchers warn that these systems still rely on outdated teaching models that focus more on memorization than true understanding.
According to a Gallup survey conducted in September 2025, around 60% of American teachers already use artificial intelligence in their daily work, primarily for lesson planning and preparation. However, the Massachusetts team points out that AI chatbots were never designed for educators, and their current applications in education reveal fundamental weaknesses.
The study analyzed 311 lesson plans generated by leading AI systems across 2,230 tasks in eighth-grade social studies. The team applied Bloom’s Taxonomy—a framework that measures levels of cognitive skills—and found that 90% of the assignments encouraged only lower-order thinking such as remembering, understanding, and applying facts. Very few activities promoted higher-order skills like analysis, evaluation, or creativity, which are crucial for developing critical thinkers.
Researchers also examined the lessons through the Banks’ four-level model of multicultural content integration, a framework used to assess inclusivity in educational materials. The findings revealed that only 6% of AI-generated lessons included multicultural perspectives, and those that did often focused on superficial elements such as holidays or heroes. The majority of the plans ignored the voices of women, African Americans, Latinx communities, Asian Pacific Islanders, people with disabilities, and other marginalized groups.
Essentially, the study concluded that AI-generated lesson plans are dull, traditional, and uninspiring. The researchers noted that chatbots, at their core, are predictive text models—machines trained to generate the next most likely word based on patterns in existing data. As a result, they tend to produce generic, formulaic teaching materials that lack creativity and emotional depth.
The researchers suggest that educators can still use AI tools strategically, not to automate their teaching but to stimulate new ideas or refine existing lesson structures. For example, rather than asking AI to “create a lesson plan on the Constitutional Convention,” teachers can provide more detailed prompts, including specific learning goals, educational methods, and critical thinking objectives.
For instance, an effective prompt might be: “Create a lesson plan for the Constitutional Convention for 8th-grade students in Massachusetts that includes at least three activities focused on evaluation or creation based on Bloom’s Taxonomy. Integrate underrepresented historical perspectives and social action projects following Banks’ multicultural framework.” Such detailed instructions allow AI to generate more meaningful and inclusive educational materials.
Conclusion:
While AI continues to transform education, this study is a wake-up call about its current limitations in fostering intellectual growth. Without human guidance, empathy, and pedagogical expertise, AI remains a powerful but mechanical assistant, not a substitute for real teaching. The future of education, the researchers argue, lies in collaboration between human educators and AI, combining creativity, cultural awareness, and critical inquiry to shape truly engaging learning experiences.
