Widespread availability of generative AI tools is reshaping how students approach exams, papers, and projects—requiring thoughtful changes to course assessment design. Traditional assessments, particularly those relying on rote recall or standard essay formats, may no longer effectively measure student learning when AI can generate plausible responses that are difficult to differentiate from human written content.
This section will focus on two aspects that instructors need to address to effectively accommodate GenAI in their assessments: academic integrity and assessment redesign.
Academic Integrity in the Age of AI
The bad news is that there are currently no reliable AI detectors and there will likely be none in the near future. This means that the best ways of preventing AI-enabled cheating are design based. We list several of these in the next section. Beyond rethinking assessments, there are relatively simple approaches you can take to address inappropriate use of AI in your classroom.
One is to set clear policies about acceptable AI use and to openly discuss the ethical dimensions of these tools with students. You will likely need to have AI use policies for each assignment and for the course as a whole. Some faculty have found it useful to engage students in shaping the GenAI course policy, which increases student buy-in and the likelihood that they’ll follow the policies. See the “Prepare Students” section in this resource for more.
You might also consider changing the assessment mix in your course to include a combination of low stakes and high stakes assessments. If a large part of a student’s grade depends on one or two products, they feel more pressure to succeed on those, which might lead them to cheat. See “Reduce Student Anxiety” in this resource for more.
Many faculty have changed the type of assessments they use in response to GenAI, moving from take-home exams, for example, to holding exams in a controlled environment with a proctor. While this approach may address academic dishonesty in some contexts, it raises other problems and is not a long term solution to the problems posed by GenAI. A much better approach is to redesign (or replace) your assessments with ones intentionally created with AI in mind.
Assessment Redesign
Many faculty are now considering how to redesign assessments to accommodate the threats and promises of GenAI. If you are in the process of rethinking how you measure student learning, consider the following:
- Identify the learning outcomes you want your assessments to measure and make sure that any new assessments you introduce align with them. Reviewing a resource such as the “AI Assessment Scale” might help with this task.
- Redesign assessments to emphasize “human skills” such as critical thinking, ethical reasoning, creativity, evaluation, or using emotional intelligence.
- Create assessments that are hard for AI to complete. These include hands-on activities (field research or observational reports), personal reflections, synchronous activities (oral reports or presentations), learning journals, and activities that include multiple kinds of media (image, text, audio, and tactile elements).
- Resign assessments to focus on process over product. Instructors might require students to submit drafts, reflections, or annotated bibliographies that document their learning journey.
- To support process-oriented assessment, It’s a good idea to scaffold large papers or projects over the semester, with due dates and feedback at various points. Portfolios can also be useful tools for measuring and assessing process.
- Motivate students to complete assignments without GenAI. Make the value of the activity clear by explaining to students why you are asking them to do it and how it will benefit them now and in the future. A useful approach here is to consider adopting “authentic assessments” that mirror real-world activities students will complete after leaving the university.
- Design assignments that incorporate AI use transparently, asking students to critique or improve AI-generated content.
Ultimately, instructors are encouraged to view AI not just as a threat, but as a catalyst for more engaging, authentic assessments that prepare students for a future in which AI will be an integral part of many professional and intellectual tasks.