We cannot fully anticipate how GenAI will impact every course or assignment at U-M. We will need to approach GenAI with an experimental mindset, draw on what we know, prioritize high-impact pedagogical strategies, and plan for incremental, sustainable instructional development. High-impact pedagogy includes collaborative learning, interactive learning, higher-order thinking, service- or community-based learning, research, authentic learning and assessment, and writing-intensive courses. Writing-intensive courses present particular benefits, opportunities, threats, and risks.
To begin this process, we recommend that instructors practice using GenAI tools relevant to their disciplines, interests, and pedagogies—before the start of fall term.
In preparation for fall term, instructors should evaluate several aspects of their courses:
- What are the course objectives and rationale for them? Can GenAI be used to meet any of the objectives?
- Are there new learning objectives in the areas of knowledge, skills, or values about GenAI that students need to meet? Will students have equitable access to GenAI for these objectives?
- What tasks do students need to complete to demonstrate they meet the learning objectives?
- How will learning be assessed?
- Do the objectives, tasks, or assessments present problems with respect to equity, inclusion, diversity, or accessibility? If so, can they be adjusted to ensure fairness and inclusion?
The course design should include some parts that cannot be completed satisfactorily (solely) by GenAI tools, unless GenAI skills and values are the primary learning objectives. GenAI offers us an opportunity to rethink learning objectives as well as assessment methods. Several principles of effective pedagogy should guide us in adjusting courses to GenAI environments:
- Larger component of technology-free in-class assignments
Including low-stakes writing, programming logic, etc. may be possible, while recognizing that in-class assignments can create inequities for some students with disabilities and English-language learners, unless appropriate accommodations are offered. - Focus on higher-order thinking
Require students to develop their capacity for creating, evaluating, analyzing, and applying (what they learned to new contexts or problems) - Emphasize the learning process over the product when appropriate
Assign scaffolded tasks developed through several stages (low-stakes writing, outlines, proposals, outlines, drafts, peer review, revision) - Prioritize authentic instruction and assessment
Assign tasks that address real-world problems or simulate real-world tasks; require students to use disciplinary theories and methods in innovative ways; ask students to address complex questions that have no easy or right answer; provide formative feedback that improves students’ performance - Require accurate and verifiable citation of sources
Have students include links to the sources they use in a project (GenAI cannot reliably do this) - Add metacognitive exercises to assignments
Assign reflections about how formative feedback is understood and what was revised in response to feedback - Teach academic integrity
In designing assignments, we should explain the purpose of the task to foster motivated learning. The purpose should align with learning objectives in the areas of knowledge, skills and values:- what do students need to learn and why—what the value of the learning is for them now and/or in the future?
- What skills will students gain from using AI, what knowledge they will need to apply.
Instructors are encouraged to try completing assignments using GenAI tools before distributing assignments to students. GenAI output can productively be used in class as a prompt for discussion about what technology does well and not so well, as well as the potential social risks and harms implicit in the output (e.g., racial bias, misinformation).
The following decision tree may be used to evaluate assignments in relation to GenAI environments:
© 2023 by University of Michigan is licensed under Attribution-NonCommercial-ShareAlike 4.0 International.
Instructors may find it useful to consider the principle of “transparent assignment design” as they review the relationship between course learning objectives and assignments. A transparent assignment template created by Mary-Ann Winkelmes at the University of Illinois offers a concise and concrete example.
Instructors may decide to assess learning with a wider range of assignments than they employed before sophisticated GenAI tools became widely available. The following strategies may prove useful.
Strategies for Engagement and Assessment
In line with best practices for teaching and learning, strategies for engaging and assessing students beyond recall and procedural, superficial understanding are effective pedagogical measures for rising above GenAI tools. Diverse teaching strategies align with diverse approaches to assessment. The following chart offers suggestions for instructors to consider as they redesign courses for GenAI environments.
| Pedagogical Strategy | Possible Assessment Methods |
|---|---|
| Active, experiential, and project-based learning sequences similar to those encountered in the discipline | In-class polls, products of peer collaboration, self-assessment |
| Flipped classrooms to develop interactive learning, peer collaboration, and problem solving skills | In-class polls, products of peer collaboration |
| Reflections on learning and research processes | Assign a description of research methods, evaluation of sources, explanation of how sources contributed to students’ thinking, and discussion of problems encountered and solved in the process. Evaluate on the basis of completeness. |
| Live debates, oral presentations, role playing | Evaluation rubric co-created with students before the performances. Evaluate on the basis of completeness. |
| Multimodal compositions, such as infographics, podcasts, drawings, websites, videos, alternative text creation for videos and images | Evaluation rubric co-created with students during scaffolded composition process Authentic group, individual, and self-assessments that require critical and creative thinking |
| Social annotation of reading | Peer review of annotation quality |
| Requirements for students to contextualize information, make connections in written work to their lives, current events, and earlier course concepts or lectures, and/or incorporate required readings, videos, images, case studies, or datasets | Evaluate on the basis of completeness and successful integration/synthesis of course materials, concepts, theories, or methods |
| Poster displays and discussions | Peer review according to a rubric co-created with students during the learning process |
Grading standards should be explicit when assignments are distributed and should prioritize higher-order thinking, veracity, critical and creative thinking, and learning processes over mechanical or grammatical correctness. The rationale for these priorities is that developing the human capacity for sentience, creativity, and reasoning aligns with the social value of higher education, while “correctness” (within narrow limits) can increasingly be accomplished by machines.
Integrating AI tools into Course Design
Teaching in an AI-augmented world will demand a reevaluation of the course design and pedagogical approaches. AI can be incorporated into a course in several ways. Examples include providing personalized learning experiences, predictive analytics to identify student progress towards learning goals, and automated grading systems. While integrating AI, it is important to critically examine:
- Appropriate use of AI
Ensure that AI is used in a way that supports the learning objectives of the course. - Necessity of AI
While AI can improve various aspects of teaching and learning, it is crucial to discern if its implementation is necessary for a particular task or if a non-technological approach would be equally or more effective. - Pros and Cons of AI
Understand the benefits, such as increased efficiency and personalized learning, against the potential risks and drawbacks such as privacy issues, technology dependency, or reducing human interaction. - Avoiding the Tail Wagging the Dog Scenario
It is essential to ensure that AI serves as a tool to achieve educational objectives and not the other way round where the curriculum is altered to fit the AI tools. - Book Guide - Using Chatbots in Classroom Assignments and Activities
- A TTC-created guide that highlights ideas for using chatbots in classroom assignments and activities. The guide is based on the ChatGPT Assignments to Use in Your Classroom Today book from the University of Central Florida.
Skills and Competencies for an AI-Augmented World
The future workforce will necessitate a unique blend of skills and competencies to thrive in an AI-driven environment. Accordingly, regardless of the field of study students should be exposed to the following aspects of GenAI:
- Digital Literacy
Students should understand how AI systems work and appreciate the intricacies of machine learning, data analysis, and algorithm design. - Ethical Understanding
As AI systems become more complex, issues around privacy, bias, and ethics will be prominent. Students need to be aware of these ethical considerations and equipped with the skills to make informed decisions based on their values. - Critical Thinking and Problem Solving
While AI can automate many tasks, the ability to think critically and solve complex problems is something that cannot be replicated in the near future by machines. The cognitive dimension of learning is more important than ever, given the necessity of evaluating, fact-checking, and verifying AI output, as well as protecting against and mitigating the risks and harms of AI. These will remain key skills in the near-future workforce. - Adaptability and Lifelong Learning
AI is a rapidly evolving field. Therefore, the ability to adapt to new technologies and a commitment to lifelong learning will be crucial.
In conclusion, while AI can greatly augment the teaching and learning process, it is critical to maintain the delicate balance between leveraging technology and nurturing the unique human skills that AI cannot replicate. Doing so can prepare our students for an AI-augmented world without compromising on the essence and social value of education.
GAI Assignment Decision Tree
- Does the assignment address a course competency or essential learning objective?
- No: Consider
- dropping
- offering as ungraded practice
- turning into class activity
- Yes:
- No: Consider
- Can AI complete the assignment and pass?
- No: Carry on, nothing to change
- Yes:
- Would letting students use AI undermine their learning and hinder your ability to monitor or evaluate it?
- No
- Integrate: Build AI into the assignment
- Yes:
- Invigilate: Have students demonstrate their learning via in-class activities, exams, etc. where AI use can be monitored.
- Mitigate: Flip your class, use alternative assignment types, leverage instructional technology tools.
- No