Navigating AI: Strategic Foundations for Your Course

Clearly defining how students should use AI, creating transparent policies, and modeling responsible AI engagement will promote academic integrity and authentic learning.

1. Set Learning Goals: As you would with any learning experience, it is important to begin with the end in mind. What knowledge and skills do you want your students to acquire? Be sure learning objectives are observable, measurable, and written in language your students will understand. Additionally, consider how AI impacts each objective. Sharing or co-creating these with your students helps them understand why they are putting effort into learning.

2. Communicate AI Policies Early: By providing clear expectations, you set students up for success. Drawing on guidelines and examples in this AI Syllabus Statement Guide, ensure all syllabi include a course-level AI policy that clearly outlines when and how to use generative AI tools, along with a supporting rationale. The AI Assessment Scale can also help you develop transparent policies. This flexible student-facing framework allows instructors to indicate the level of AI use permitted in an assignment and lets students know what that entails. Additionally, if an assignment requires AI, consider options for students who may oppose AI use.

3. Model Appropriate AI Use: Demonstrate how AI tools can benefit tasks such as idea generation and reviewing literature. As a disciplinary expert, model effective AI use so students gain insight into the technology’s benefits and limitations, as well as its alignment with academic integrity policies. Include an AI transparency statement that describes how you used AI tools to create specific content, and require students to include a similar statement with their work. Consider using the Pause Before You Prompt framework to model the thoughtful reflection on the intentions, uses, and impacts of AI tools.

In the age of AI, developing uniquely human skills has never been more important. Prioritizing learning experiences that foster abilities AI can not easily replicate strengthens human connections and critical thinking.

1. Keep Humans in the Loop: Encourage students to begin projects by thinking, reflecting, empathizing, drafting, and meeting with peers before engaging with AI tools for feedback or enhancements. Similarly, after using AI tools, students should verify the accuracy and appropriateness of any AI output they include in their work.

2. Build Community: Design learning experiences that build community through student-to-student interaction and student-to-teacher interaction. These interactions build trust, setting students up for success and reducing the likelihood they will seek AI-powered shortcuts.

3. Model and Encourage Critical Thinking: Treat AI tools as fallible learning aids, not as definitive experts. Students need to know how to evaluate AI output, not just how to prompt the tools. Explore the ethical issues surrounding AI with students, including algorithmic bias, environmental impacts, data privacy, and inequitable access to resources. Consider assigning tasks that require students to critique AI outputs using discipline-specific frameworks, and prompt them to ask who is missing or what stereotypes are being reinforced.

4. Prioritize Human Skills: Cultivate human abilities such as empathy, creativity, ethical reasoning, and interpersonal communication. For example, asking students to present a live or recorded summary of their project will highlight their thinking and communication skills. Similarly, assigning students to interview an expert in the field will help them practice interpersonal skills and connect the course materials to the real world.

Learning happens when students grapple with complex ideas. Instructional strategies should ensure learners remain actively engaged in the learning process.

1. Embrace Desirable Difficulty within Limits: Struggling with content activates the brain’s natural learning mechanisms; however, too much struggle can prevent the formation of strong mental models. In practice, this means instructors might encourage learners to generate answers and wrestle with problems before turning to generative AI. Furthermore, instructors should not let students flounder indefinitely without support. Finding the right balance helps students to build confidence and retain knowledge.

2. Design Motivating Real-World Tasks: Assign tasks that are interesting and related to your students’ professional and discipline-specific work. For example, in a public affairs course, this may be a project proposal presentation to the city council. In a marketing course, an authentic summative assessment may include a collection of marketing artifacts that each demonstrate a theory of motivation. Authentic assignments can improve students’ engagement, satisfaction, and employability, reducing the likelihood of offloading tasks to AI.

3. Integrate AI as a Supplement, not a Substitute: Research indicates that learning improves when AI is used under guidance. In practice, this may mean encouraging students to use AI tools that ask questions or prompt reflection, rather than simply providing answers. AI technology should support cognitive effort, not replace it.

4. Emphasize Processes Over Products: Design assessments and activities that highlight the learning path over the outcome, as AI can easily generate passable products. In practice, instructors can break larger projects into smaller, targeted tasks that allow for timely, specific feedback. Similarly, students may be required to submit several drafts of a project over the term, each with brief notes explaining how they responded to instructor, peer, or AI feedback. These approaches can reveal student thinking and reward genuine growth.

5. Make Time for Reflection: Require students to reflect on their learning process – with and without AI. This supports metacognition — or thinking about your own thinking — and helps students to consider their challenges, areas of growth, and self-regulation strategies. In practice, this might mean assigning cognitive (exam) wrappers, targeted reflections, or process journals. Students with a good understanding of their own cognitive processes are less likely to develop a need to shortcut their learning.

While AI detection tools remain unreliable and biased toward non-native English writers, instructors must use assessment strategies that align with AI realities to measure authentic student learning.

1. Incorporate Oral Assessments and Multi-Modality:  Design assignments that require students to present information verbally or work in multiple modes, including text, images, slides, video, and audio. This may also appear in multiple stages, for example, an oral assessment after a multiple-choice test. Not only does this highlight our human strengths, but it also prepares students to work in dynamic, technology-rich environments.

2. Include Equitable Assessment Practices: Students are less likely to outsource their thinking to AI when assessments are designed with equity and flexibility in mind. Clear rubrics communicate expectations, helping students to feel more confident about completing the assignment, and ensuring instructors grade more consistently. Ensure students can complete AI-required assignments using free versions so as not to unfairly advantage students with paid subscriptions. Offering students a choice in assignment topics and opportunities for revision gives them more reason to value their own work over AI-generated content.

3. Foster Human-AI Collaboration: The line between human and AI decision-making will continue to blur. It is important that students learn to use these tools effectively and ethically. Design authentic tasks that assume or encourage AI use, preparing students for work environments where AI use is essential, not optional.

4. Use Controlled Assessment Environments Only When Needed: Reserve in-person or proctored environments for sensitive, high-stakes assessments. Prioritize this approach when it is the most effective way to measure your learning objectives.

5. Adopt a Multi-Lane Approach: Rather than attempting to out-design generative AI, instead acknowledge the realities of an AI-saturated world. Originally proposed by scholars at the University of Sydney, the ‘two-lane’ approach suggests that assessments either occur in a secure environment without AI (Lane 1) or in unsupervised human-AI collaborations (Lane 2). Others advocate adding a third middle lane with limited AI or expanding to a six-lane approach, where each lane signals the clear purpose of AI within an assessment. Adapting these approaches can improve clarity for teachers and learners while balancing the assurance of learning with AI realities.