To access a PDF version, click here.

Why it matters

  • STEM education has traditionally been characterized by content- and instructorcentered teaching, but it has increasingly shifted toward active learning approaches (problem-based, project-based, inquiry-based, and collaborative) that build higherorder thinking skills.
  • Generative AI (GenAI) is becoming an integral part of STEM education because it offers new opportunities to enhance active learning rather than simply automate tasks.
  • GenAI tools are now widely used by STEM students, but most applications remain individual rather than collaborative, even though active learning often emphasizes collaborative knowledge building.
  • Understanding how GenAI is being used in active learning can help instructors integrate it more intentionally and effectively. 

Purpose and context

The study aimed to:
  • Examine the distinct ways GenAI is used to support human interaction in active learning.
  • Identify and map how these interactions support higher-order thinking skills.

This review synthesized current studies examining GenAI use in active learning across STEM education, from high school through graduate education.

Key findings

  •  From the studies reviewed, five categories of human-GenAI interaction emerged: Tutoring, Co-creating, Processing, Coaching, and Simulating.
  • GenAI was most frequently used to support problem-solving, critical thinking, computational thinking, creative thinking, and metacognition.
  • Benefits of GenAI use included on-demand support, immediate and personalized feedback, and increased cognitive capacity for higher-order tasks when GenAI handled more routine work (e.g., writing code, retrieving information).
  • Overreliance and the need for students to develop prompt and AI literacy skills were major concerns; students and researchers alike worried that relying on GenAI for answers could undermine critical thinking, creative thinking, and problemsolving skills.
  • GenAI applications were predominantly individual in scope (N = 26 studies) rather than collaborative (N = 16), leaving the potential for GenAI-supported collaborative knowledge building largely underexplored. 

Recommendations for practice

  •  Use GenAI to scaffold higher-order thinking by providing prompting guides, guiding questions, or structured rubrics, rather than assuming students will use GenAI productively on their own.
  • Design learning activities that incorporate reflection, such as journals or peer critique of AI-generated content, to mitigate overreliance and keep students actively evaluating GenAI output rather than accepting it uncritically. 
  • Explore collaborative uses of GenAI, such as group problem-solving and peer review of AI output, since most current practice is individual and this area remains largely unexplored.
  • Match task complexity to learning goals; use simple problems to build student confidence with GenAI before introducing complex, open-ended tasks that demand deeper higher-order thinking.
  • Encourage productive struggle by adopting guidance-based tools (e.g., hints and prompts that help students formulate their own answers) rather than tools that simply generate answers. 
References

Otto, S., Lavi, R., & Bertel, L.B. (2025). Human-GenAI interaction for active learning in STEM education: State-of-the-art and future directions. Computers & Education, 239:105444. https://doiorg.libproxy.library.unt.edu/10.1016/j.compedu.2025.105444