AI in Teaching: Focus and Adapt Teaching for AI

Often, major new technologies and societal shifts force us, as educators, to adapt by revisiting fundamentals:

Clearly defining the most important knowledge and skills students need, both today and in the future.
Carefully framing the learning objectives students need to accomplish during the course, relative to the skills they need long term.
Clearly communicating to students the relevance of these skills and objectives and obtaining their buy-in to the learning process.
Designing and developing authentic assessments that measure students’ abilities and allow the students and teacher to identify what the students have learned.
Designing and developing content and experiences that help students learn and practice important concepts.
Providing timely and constructive feedback and helping students stay on track and focused on course objectives.

This page explores how these fundamentals apply to adapting our teaching to AI.

Defining an Objective-Driven Approach to AI

The extent to which students are encouraged or discouraged to use AI should begin with a clear understanding of the course objectives. In some courses, reliance on AI tools may prevent students from developing foundational skills the course focuses on, such as writing, language, or computational skills. In other courses, AI use will not prevent learning objective achievement and may actually prove beneficial. For example, in a strategy-oriented class, AI may extend students' ability to bring ambitious project ideas or campaigns to fruition in a short amount of time.

A course may also have a mix of learning objectives, some of which are negatively impacted by AI use while others are not. For example, students may need to develop fundamental skills early in a course, but could use AI later to save time for working on advanced skills.

Understanding the course objectives is the most important starting point for adapting course design to AI.

Designing Assignments that Maximize Student Contribution

Whether you design an assignment that prohibits or permits AI use, you face the same challenge: requiring more of students than the capabilities of AI tools, alone, can accomplish. Here are some strategies for crafting assignments that focus on students' unique perspectives, critical thinking, and problem-solving skills:

Communicate openly about AI. Acknowledge the strengths and limitations of AI relative to the topic of the course and the nature of its assignments. Consider and acknowledge the various ways students may use AI, such as research, feedback, help with task management, and more. Clearly identify in the syllabus and assignment descriptions the acceptable and unacceptable uses of AI for your course.
Emphasize thoughtfulness, analysis, synthesis, creativity, and evaluation. Avoid asking for simple descriptions and restatement of information. Case studies, applied problems, and reflective essays are examples. Note, however, that AI chatbots are steadily improving at these sorts of tasks. Although they tend to simply restate commonly published ideas and arguments, undergraduate college students tend to do the same, regardless of the assignment prompts. Additional techniques may be needed.
Use staged assignments to focus on process. Rather than evaluating a single, finished product, break an assignment into stages and have students turn in drafts, showing the evolution of their process and understanding. If permitting AI use, have students indicate what role AI played at each stage in the process.
Use layered assignments that build upon each other and achieve complex solutions. Similar to the suggestion above, focus a series of assignments on a comprehensive product or solution that uses multiple artifacts that build on each other.
Ask students to describe and reflect upon their processes. Encouraging student reflection (or metacognition) on the learning process is an effective instructional process as is. It has the added benefit of discouraging undisclosed reliance on AI.
Incorporate class-specific content. Have students reference in-class discussions, examples, case studies, and experiences that AI would not be able to reference.
Address localized and real-world problems. Generally speaking, the less abstract a problem or assignment is, the less capable AI will be at producing a truly usable solution.
Use newer, less-used sources. Older and more heavily used books and articles will have far more commentary available online for AI tools to draw from. The same AI tools will have less to work with for newer, more obscure sources. They will still produce output in response to prompts, but that output will more likely be inaccurate or off the mark.
Use a mix of assessment types. Do not rely entirely on writing or homework problems. Find other ways for students to demonstrate their competency, including presentations, videos, oral interviews, in-class exercises, and more. This approach tends to be a more universally inclusive and comprehensive form of assessment anyway.
Try to use AI to complete your own assignments. If you can adequately complete your assignments using AI, students will be able to do so as well. As you try to complete your own assignments using AI, you will hopefully identify ways you can make your assignments less AI-friendly. You will also get a good sense of what an AI response to your assignment looks like so that you can better recognize when students are over-relying on AI.

Designing Assignments that Incorporate AI

Encouraging students to maximize their contribution to an assignment is important, but students are entering a world where they will most likely be expected to use generative AI in some form as professionals. Course assignments can be an opportunity for students to learn to use AI effectively and appropriately. AI can also be used as a teaching aid to assist with feedback, simulation and modeling, research assistance, and more. Here are some ideas:

Assign students to critically evaluate AI output. In classes where research, accuracy, communication, argument, and targeted outcomes are important, having students use AI and evaluate its effectiveness can help them identify where it falls short and understand the need for domain knowledge and critical review.
Encourage AI for prototyping. Generative AI is capable of producing images, code, audio, video, and large amounts of written copy that students could only previously conceptualize in a semester's timeframe, rather than actually produce. Students can now be encouraged to use generative AI to produce functional prototypes of their ideas within the same timeframe.
Use AI for simulation and modeling. If you can engineer (or help students engineer) the right prompts, you can use AI to generate interactive case studies and simulated scenarios or conversations. USU departments, researchers, and educational technology vendors are busy developing and piloting ways to build AI-guided activities into applications that integrate with course content and learning management systems like Canvas.
Guide students in using AI for feedback. AI can provide a certain level of feedback on student work, be it written, coded, or calculated. Help students learn to use AI as a way of evaluating and improving their work.
Encourage AI-assisted research and problem-solving. Assign students to use AI as a research assistant, a tool for data analysis and visualization, debugging, and more.
Help students use AI for brainstorming. Often the hardest part of writing is generating ideas and getting started. Allowing students to use AI to generate ideas and initial outline drafts can significantly aid in the brainstorming and prewriting process.
Teach prompt engineering. Design assignments that specifically focus on teaching good prompt engineering. Grade students on the quality of their prompts and the quality of the output.

These are just a few starting ideas. Be creative. There are many other possibilities for AI to play a useful role in the teaching and assessment process.