海角社区

Generative AI tools change what students can produce quickly. The goal is not to create 鈥淎I-proof鈥� assessments. The goal is to design assessments that still measure student learning and to respond to integrity concerns in ways that are fair, evidence-based, and defensible.

The guidance on this page draws from 海角社区鈥檚 broader Authentic Assessment resources and represents selected highlights rather than the full framework. Faculty are encouraged to review and leverage the complete documents alongside this page when designing or revising assessments.

This guidance supports faculty in designing, delivering, and evaluating assessments in an educational environment shaped by widespread access to generative AI. It is grounded in a simple expectation: student work must truthfully reflect the student鈥檚 own knowledge and skills, regardless of modality or assignment type.

• Academic Integrity: Assessments must uphold standards of honesty. When generative AI is used in ways that violate stated course expectations, it may be addressed as academic misconduct under 海角社区鈥檚 Student Code of Conduct.
• Transparency: Expectations for authorized and unauthorized AI use must be communicated clearly in syllabi and assignment instructions.
• Andragogical Alignment: Assessments should measure intended learning outcomes and emphasize student-generated results aligned to course objectives.
• Equity and Fairness: Standards should be consistent for all students, and any allowed tools should be equally accessible.

Artificial intelligence is a tool, and its educational impact depends on alignment with learning goals and ethical boundaries. When AI use is addressed only through broad bans without alignment to learning objectives, students are incentivized to focus on circumventing restrictions rather than demonstrating genuine understanding and skill development.

A more effective approach is to define and communicate an ethical framework for AI use in coursework and assessment, emphasizing transparency, accountability, and human oversight. Alignment between course goals, assessment design, and AI expectations allows these tools to strengthen learning rather than undermine it.

The guidelines below focus on designing assessments that measure student learning in an AI-present environment rather than what automated tools can produce.

Backward Design: Outcomes First

Backward design begins by defining what students should know and be able to do by the end of the course, then designing assessments to measure those outcomes, and then selecting learning activities that prepare students for success. In an AI-present environment, backward design helps avoid generic or recall-based assessments that can be completed easily with common tools and shifts emphasis toward reasoning, creativity, and critical thinking.

For a concise overview of backward design, see from Ohio State University recommended by our Center for Teaching, Learning, and Design (CTLD).

AI-Resistant Tasks

While not AI-proof, AI-resistant tasks still require students to think, apply, and explain in ways that are difficult to outsource.

• Emphasize process over product (notes, outlines, drafts, revisions, checkpoints).

• Use personalized, contextual, or experiential prompts tied to student context or recent events.

• Incorporate real-world and interdisciplinary problem solving that requires synthesis and justification for a realistic audience.

Authentic Assessment

Authentic assessment places students in professional, civic, or scholarly contexts where they apply what they are learning.

• Use collaborative projects, case studies, and scenario-based evaluations.

• Require explanations of reasoning and decision-making, including tradeoffs and rejected alternatives.

The strategies below can be combined and adapted across disciplines, modalities, and AI use approaches (Allowed, Mixed, Restricted).

These guidelines apply across modalities, but asynchronous online courses require additional design attention because students complete all work outside scheduled class meetings. In asynchronous environments, expectations must be communicated clearly in writing and reinforced consistently in Canvas.

Recommended moves in asynchronous courses:

• Make AI expectations highly visible in Canvas (syllabus, Start Here module, and every major assignment or quiz).
• Use asynchronous versions of oral checks (short recorded explanations that can be sampled or spot-checked).
• Build time-bound, individualized quizzes and knowledge checks using question banks and randomization when appropriate.
• Leverage process documentation and staged submissions through Canvas for major projects.
• Use online discussions and reflections strategically, and state AI expectations explicitly with a brief rationale.