Building an Understanding of AI in Learning Environments

Mike Hassett, Director of Platform and Learning Technologies, Western Governors University

Mike Hassett, Director of Platform and Learning Technologies, Western Governors University

Mike Hassett with over 20 years of experience is bridging technology and learning. He transforms strategy into scalable digital ecosystems that advance educational access and quality. As a proven innovator and leader, Hassett focuses on building platforms that empower learners, streamline operations and shape the future of online education.

If your institution is like mine, it is being inundated with requests to either restrict the use of AI or implement the use of AI broadly across learning and assessment. One of the problems we face in these conversations is what exactly “AI” means to the person or people making the request. Without that understanding, it is difficult to know how to proceed, especially in an onslaught of seemingly conflicting requests. To clarify these conversations with staff and faculty, I have created a taxonomy of student AI uses for learning and assessment.

AI experience type	Sample use case	Concerns to be aware of
Support Learning
AI Exposure - Helping students become comfortable with and more adept at using AI, including prompt engineering	Students are provided instruction on effective prompt engineering, then provided with one or more AI tools in which to author prompts to evaluate responses. AI can both respond to the prompt and coach the student on prompt quality.	Requires control over tools and interaction to prevent superficial engagement and quality problems with student inputs and LLM outputs.
AI Tutor - Open-ended or directed conversations with AI bounded by a specific course or topic	Students have a university-built AI course companion that can be accessed at any time while completing a course to ask for clarification of key concepts.	AI requires access to course content and extensive training in instructional and assessment practices. It needs solid guardrails to prevent students from misusing the tool and to prevent the misuse of intellectual property.
AI Assistant - When a student is completing certain tasks, AI offers real-time help with the task	Students are given subscriptions to Grammarly, which offers real-time input on the structure, clarity and correctness of their writing.	This requires purpose-built tools that can plug into other tools being used to complete tasks, which can be expensive.
AI Collaboration - AI partners with individual students or groups of students to complete a task	Students are given a task to create a presentation and they are given access to an AI tool. They are encouraged to use the AI to assist them as they think through the various process elements to improve their final product.	This presents difficulties in evaluating student learning, clarifying the division of labor and preventing an overreliance on AI.
Evaluation
AI Evaluation - Formative and summative assessment evaluation using AI alone or with a human evaluator	Students working through a course are presented with writing activities that are evaluated entirely by AI. At the end of the course, the students’ final projects are evaluated by AI, and failing evaluations are reviewed by human evaluators as a quality check.	It is difficult to ensure the validity and reliability of AI evaluations and to prevent equity issues in evaluation responses. Ensuring validity and reliability requires a significant investment of time and money.
Practice/Application
AI Role Playing - Facilitating character-based interactions for skill development in which the AI plays one or more characters	Students in a nursing program are provided an AI-driven activity in which they conduct a health assessment of a patient and receive coaching to improve their technique.	Role playing requires very robust prompts to ensure accuracy and appropriateness of the AI outputs, particularly in high stakes or complex scenarios. It also requires close monitoring to ensure both students and AI are acting appropriately.
AI Simulation - Replicating complex environments for learning in which AI drives the behavior of the environments	Students in a network engineering program use a virtual lab in which AI provides tickets from users, students make network changes, and based on those changes, the AI delivers additional responses to the original ticket.	Robust simulations require significant investment in infrastructure to provide computational resources and complex interactive environments. They also require significant testing to ensure accurate performance in real-world contexts.
Creation
AI Development - Building AI tools, including GPTs	Students in an elementary education learning experience design program are taught how to build GPTs to use with their own students.	This requires solid guardrails to prevent misuse of the tool; this can be problematic when the goal is to expand the number of tools students might use.

The categories in this taxonomy are not mutually exclusive; courses and classrooms can use multiple AI experience types simultaneously. While not definitive or static, this taxonomy can serve as a solid foundation for investigations and conversations. Use it to clarify what stakeholders envision when they propose implementing or restricting AI in a learning or assessment experience, to prevent overlapping and redundant tools and to avoid problematic interactions by providing the appropriate AI training and guardrails.

Author’s note: I used Claude.ai (Sonnet 4.5) as a collaborator on this article. I initially had Claude review the taxonomy for uniqueness and completeness (Claude suggested adding AI Collaboration). I later had Claude provide input on the article’s structure and word choice, but I made all decisions on the final shape, content, and wording. Any mistakes are mine.