1. Home ›
  2. Resource Library

AI for Teaching and Learning | Assessment |

Guidance on the Use of Artificial Intelligence for Assessment

Author: University Committee on Academic Technologies (UCAT) AI Subcommittee

May 6, 2026

The rapid deployment of artificial intelligence (AI) tools in society at large has invited us to rethink many of our classroom practices, both around what sort of work we expect students to produce and how instructors can or should use AI to aid their efforts to assess student work and assign grades. This document lays out some general guidance from the University Committee on Academic Technologies (UCAT) AI Subcommittee on how instructors should approach assessment practices assisted by AI.

Definitions

Briefly, we clarify a few terms that are used in varied ways across disciplines.

  • Assessment is the overall process of gathering and evaluating information about student performance, and using this to guide decisions about both teaching and grading.
  • Grading is the specific matter of assigning a letter or number grade to summarize a student’s achievement of learning goals, on a unit of work or the course as a whole. 
  • Feedback is the process of providing detailed responses to student work for a variety of purposes, such as correcting errors, encouraging growth, or explaining grades.1

Framing Questions and Guidelines

The committee considered two primary questions:

  • Should instructors use AI for grading, and what are the important considerations?
  • Should instructors use AI for feedback, and what are the important considerations?

Most importantly: Instructors must remain the final decision-maker on grades.

Grading is ultimately a human activity that serves to summarize the student’s progress toward the course learning goals. While AI might be used to assist in generating observations that support grading, such as pointing out technical errors or drafting general comments, instructors retain full responsibility for grades assigned on any work and must be able to explain and justify the observations supporting that grade. AI should not be used for fully automated grading of student work; the final judgement, decision-making power, and the resulting score must reside entirely with the human instructor.

Before adopting AI tools to generate feedback or other observations about student work, instructors should carefully consider the following aspects:

Instructors should consider what is potentially lost (or gained) by adopting AI-assisted feedback.

A close reading of student work is one way that faculty develop a deeper understanding of each student’s strengths and weaknesses, as well as the performance of the class as a whole. By taking the time to digest and respond to student work thoughtfully, faculty have the opportunity to connect more closely with individual students. Consider the tradeoffs in adopting AI tools for assessment: Will an AI-assisted process create more distance between you and your students? Will it provide feedback that is not otherwise available? Will it free up more time for one-on-one engagement with students? A careful consideration of the intended student-faculty relationship should guide decisions about AI adoption.

Instructors should disclose to students if, how, and why AI is being used in a course.

Indicate whether AI is used for assignment creation, drafting feedback, guiding faculty attention, producing student-visible output, or some other purpose. If AI is not used at all, that should be clearly stated as well. This helps to preserve trust between students and instructors. Similar disclosures should be made in course planning and departmental curriculum discussions in order to help establish norms and expectations across faculty and courses. Consider the example disclosure in Appendix A based on Algahami.2

Instructors should use only tools approved by the University for handling student work.

Online AI services vary considerably in how they deal with privacy, security, and the use of input data as training material for later outputs. There are substantial risks that unapproved tools may leak personally identifiable data or cross-contaminate results from different students. Instructors should not feed, or allow access to, student data of any kind into tools not approved for that purpose. (Currently approved tools are noted in reference 3.)

Instructors should be well-versed in best practices for influencing and assessing the behavior of AI responses to student work.

Before relying on any AI-produced feedback, instructors should compare it against human feedback on a variety of samples in order to determine the quality of results, alignment with the course learning goals, and potential bias. AI models are fundamentally probabilistic and thus are prone to producing compelling but unfounded conclusions, behave in unexpected ways with new inputs, and change over time with new model versions. Methods for ongoing monitoring of feedback quality should be put in place to guard against unexpected changes in the system behavior.

Instructors should consider whether students would benefit from access to the same tools.

If an automated AI tool is useful for feedback by the instructor, it may also have formative value for the student. Providing access to the same tool gives students opportunity for growth while working independently. This also serves to reserve the instructor’s effort for higher level evaluation with a “human touch.” On such occasions, transparency and guidance are once again paramount: Model best practices and define relevant use-cases for students’ independent deployment of these tools. Consider in advance how to prepare students for the inevitable situation where AI generates feedback that is contrary to the instructor’s goals.

Instructors should consider how AI adoption impacts Teaching Assistants (TAs).

Providing detailed feedback is a task that has often been given to TAs. This has allowed the process of feedback to scale up to large classes but also provided TAs an opportunity to develop their own teaching skills for the future. Instructors should consider what new skills TAs ought to develop, and what skills might be shortchanged by offloading tasks to AI. For example, consider involving TAs in constructing AI feedback approaches and using TAs as part of the accountability team that reviews feedback given to students. Ensure that TAs have a full understanding of the acceptable uses of AI by both students and instructors.

Further Guidance

We fully expect that the AI landscape will continue to change, and instructional techniques will evolve as the academic community gains more experience and understanding. For further guidance around the use of AI in teaching and learning, instructors are invited to reach out to ND Learning.

References

  1. Marion, S. F., Pellegrino, J. W., & Berman, A. I. (Eds.). (2024). “Reimagining Balanced Assessment Systems.” National Academy of Education.
  2. Alghamdi, L. H., & Alghizzi, T. M. (2025). “Educators’ reflections on AI-automated feedback in higher education: A structured integrative review of potentials, pitfalls, and ethical dimensions.” Frontiers in Education, 10, 1704820.
  3. Approved AI Tools @ Notre Dame
  4. Notre Dame’s Responsible Use of Data & Information Technology Resources Policy

Members of the UCAT AI Subcommittee

  • Roger Alford, Law School
  • Alex Ambrose, Lab for AI in Teaching and Learning
  • John Behrens, Technology & Digital Studies Program
  • Liang Cai, Department of History
  • Roberto Cásarez, Office of Information Technology
  • Mike Chapple, Department of Information Technology, Analytics, and Operations
  • Tara Kenjockety, Hesburgh Libraries
  • David Leighton, Department of Chemical and Biomolecular Engineering
  • John Lubker, Associate Dean for Academic Affairs, Graduate School
  • Elena Mangione, Department of Romance Languages and Literatures
  • Ron Metoyer, Vice President and Associate Provost for Teaching and Learning
  • Nathaniel Myers, University Writing Program
  • Paul Perrin, Keough School of Global Affairs
  • Connor Rezner, Undergraduate Student
  • Ardea Russo, Director of Academic Standards
  • Ragnar Stroberg, Department of Physics and Astronomy
  • Kasey Swanke, Keough School of Global Affairs
  • Douglas Thain (subcommittee chair), Department of Computer Science and Engineering
  • Steven Varela, Office of Information Technology
  • Victoria Woodard, Department of Applied and Computational Mathematics and Statistics
  • Xiangliang Zhang, Department of Computer Science and Engineering 

Appendix A– Sample Disclosure of AI Use in the Syllabus

Courtesy of Alex Ambrose, CDT 20600, Learning with Generative AI

Use of AI-Assisted Feedback in This Course

Human-in-the-Loop Approach: In this course, I may utilize artificial intelligence (AI) tools to assist in providing you with feedback on your work. My goal is to leverage AI’s ability to provide immediate, tailored, and personalized guidance so that you can address gaps in your understanding while the material is still fresh in your mind.

However, I remain an “active mediator” in this process. This means that AI does not replace my role as your instructor; instead, it augments my ability to support your learning.

My Commitments to You (The FATE Framework): To ensure this technology is used responsibly, I adhere to the following four principles:

  • Fairness: I am aware that AI can sometimes reflect algorithmic bias based on the data it was trained on. I critically evaluate all AI-generated suggestions to ensure they are fair, unbiased, and inclusive of all learners.
  • Accountability: I am the ultimate arbiter of all feedback and grades. While AI may suggest comments or identify patterns, I review and approve the final output to ensure it aligns with our course's pedagogical goals.
  • Transparency: I will clearly disclose which assignments utilize AI assistance. You have the right to know how your feedback is produced.
  • Ethics and Privacy: Your data privacy is a priority. I only use tools that comply with institutional data protection standards to ensure your sensitive information remains secure.

Maintaining Our Connection: The most important part of this course is our human relationship. AI is used only for routine feedback tasks to allow me more time for one-on-one support and meaningful instructional design.

Your Role: If you ever feel that AI-assisted feedback is inaccurate, confusing, or does not reflect your effort, I encourage you to reach out to me directly. I am always available to discuss your work, override automated suggestions, and provide the human perspective that technology cannot replicate.