Empowering Instructors | Supporting Excellence
Reimagining Assessments
When redesigning assessments to ensure they truly measure students’ learning given the ubiquity of generative AI tools, it is important to keep your pedagogical values in mind. For any assignments completed outside of class, it is safe to assume students will be using generative AI tools during some point of their creative process.
Consider what knowledge students should develop by the end of the class. Are the desired outcomes focused on factual recall, critical analysis, creative synthesis, or constructing disciplinary arguments? How effectively do current assignments align with and advance these objectives? Once these outcomes are clearly articulated, instructors can more readily evaluate how generative artificial intelligence (GenAI) may disrupt or enhance students’ learning processes.
Reflect on the following questions to guide your thinking:
- To what extent is the process of writing or idea development integral to the intended learning outcomes?
- In what ways might GenAI be used to facilitate, rather than circumvent, authentic learning?
- Where might clarity about your expectations be needed?
Depending on the responses, instructors may find it valuable to redesign assignments to emphasize higher-order cognitive skills that GenAI tools cannot easily reproduce. Generative AI excels at content generation but struggles with evaluation. Consider:
- Allowing students to use generative AI to form their arguments but holding them accountable for the output’s veracity
- All sources, arguments, images, and so forth must be properly cited and the student’s grade will be affected by any inaccuracies
- Having students critique AI-generated writing and explain why human-generated arguments may be superior
The majority of students want to follow the rules. They understand the value of their degrees, and they also understand that their generation is the most apt to be “replaced by” generative AI. If you take the time to clearly explain why your generative AI usage policy is set the way that it is, most students will respect your guidelines and try their best to work within them.
Perkins et al. (2024) developed an AI Assessment Scale that has been widely adapted in higher education classrooms. As you rethink your own assessments, it is worth reviewing the scale and mapping how your current objectives align.
Consider, too, that overly restrictive generative AI policies, like requiring in-class handwritten papers, may unintentionally disadvantage students with accessibility needs or who are non-native English speakers.
Assessments are meant to measure students’ learning, but they also are an opportunity for additional exploration of a topic. It is well established that students’ retention of material is stronger when they are asked to teach the content.
- Instead of a paper, require an application as the assignment’s submission. Generative AI programs like ChatGPT’s Codex and Claude’s Code make creating applications easier for regular end users (those without software engineering background).
- Based on the lesson’s topic, accept assessments that show the student learned a new skill with Gen AI: creating a webpage, an application, a remediated PDF, or similar.
- Despite having a wealth of graphic design tools at their fingertips, students still create uninspired presentations in PowerPoint, Google Slides, or Canva.
- Assess students on both the content of the presentation and its delivery.
- Presentations should be dynamic, visual storytelling apparatuses that emphasize accessibility for audience members (and not banal bulleted slides).
- Generative AI is ubiquitous, almost perniciously so. What benefits and risks has Gen AI presented for your chosen discipline or career?
- Apply lessons from class to real-world use cases and learn how Gen AI is changing the field.
- This generation of students is at the greatest risk of being “replaced” by AI; course assignments should emphasize students’ relevance
For those looking to mitigate students’ increasing dependency on Gen AI tools, require assignments that thoughtfully integrate in-class experiences: references to specific class discussions, presentations, lab work, in-class active learning exercises, and so forth. The Gen AI tool will not have this knowledge so students must apply classwork to assessments unassisted.
Often, faculty include a lot of information in their syllabus about how students are or are not allowed to use generative AI in their course.
However, because AI agents can navigate into your UB Learns course, you are advised to include a statement regarding what students can or cannot do with your intellectual property: PowerPoints, lecture notes, images, handouts, assignments, and so forth.
“Course materials, such as lecture notes, slides, and exams, are protected by copyright, and students are prohibited from reproducing or distributing them without the professor’s explicit written consent. Students may use the materials for their own educational use, but unauthorized distribution, including uploading materials to a generative AI tool like ChatGPT or Google’s Notebook LM, is a violation of copyright and the university’s academic integrity policy.”
Looking to monitor students’ use of generative AI tools when drafting their course papers? Consider requiring students compose their papers in a cloud-based word processing software tool, like MS Word online or Google Docs. Cloud-based word processing software programs have revision histories available. If there was ever a question of originality, students can show you the revision history of their paper as it was drafted over time. Students receive licenses to Microsoft 365 tools with their tuition, so all have access to MS Word online.