University of Vermont AI Policy

There are varied ways artificial intelligence tools can be used in academic work. For students, the most important question to ask in a course context is “what has my instructor said about the ways AI tools can be used in this class or assignment?” Expectations and opportunities will differ from class to class, as each set of assignments is carefully designed toward particular learning goals. Students should talk with their instructors, pay attention to policies on the syllabus, and note how assignment instructions reference technology use.

Students must author their own work and cannot use generative AI tools to complete graded items (including but not limited to readiness assurance tests and summative exams) unless explicitly permitted by faculty.

In any assignment, students must disclose their use of generative AI tools in completion of assignments including reflections and case reports, specifying tools used. Failure to do so is against the Code of Academic Integrity and will be responded to in accordance with the Enforcement of University and Larner Behavioral Standards Policy.

There are varied ways artificial intelligence tools can be used in academic work. For students, the most important question to ask in a course context is “what has my instructor said about the ways AI tools can be used in this class or assignment?” Expectations and opportunities will differ from class to class, as each set of assignments is carefully designed toward particular learning goals. Students should talk with their instructors, pay attention to policies on the syllabus, and note how assignment instructions reference technology use.

Students must author their own work and cannot use generative AI tools to complete graded items (including but not limited to readiness assurance tests and summative exams) unless explicitly permitted by faculty.

There are varied ways artificial intelligence tools can be used in academic work. For students, the most important question to ask in a course context is “what has my instructor said about the ways AI tools can be used in this class or assignment?” Expectations and opportunities will differ from class to class, as each set of assignments is carefully designed toward particular learning goals. Students should talk with their instructors, pay attention to policies on the syllabus, and note how assignment instructions reference technology use.

Use of AI in developing self-study and self-assessment tools distinct from graded assessments is permitted

There are varied ways artificial intelligence tools can be used in academic work. For students, the most important question to ask in a course context is “what has my instructor said about the ways AI tools can be used in this class or assignment?” Expectations and opportunities will differ from class to class, as each set of assignments is carefully designed toward particular learning goals. Students should talk with their instructors, pay attention to policies on the syllabus, and note how assignment instructions reference technology use.

Broad adoption of generative AI tools across many domains has had a swift and often unclear impact on research and innovation. UVM provides a roadmap of resources for researchers who want to ethically engage with generative AI tools in our research and innovation ecosystem.

When submitting scholarly work for publication or presentation, AI use must be reviewed and approved by the faculty advisor, adhere to requirements for individual publication or venue and must be disclosed. Disclosure must include the specific AI tool and how it was used.

High-risk data such as UVMHN protected health information (PHI), personally identifiable information (PII), non-public protected data (NPPD), or sensitive research data can never be entered into public domain AI solutions, such as Chat GPT, Bard, Bing, etc.

The University of Vermont has a contract in place with Microsoft Copilot but does not have a contract or agreement with any other AI tools or services. This means that standard UVM security, privacy, and compliance provisions are not in place when using technologies, other than Copilot.

If researchers want to use an AI tool with data determined to be moderate or high-risk/restricted, the AI tool is subject to applicable university policies around contracted purchasing, including an ISO Review.

If a researcher is looking to use a medical device, software or computer system using AI that will interact with UVM Health System servers, EPIC or within the network firewall, prior approval from IT will be needed before starting the study.

Artificial Intelligence (AI) is an ever-evolving field that presents new opportunities for research while amplifying the risk of data privacy and confidentiality. This guidance covers the use of AI, machine learning, deep learning, and related AI techniques used in research activities.

AI cannot obtain automatic informed consent from participants; that is, a human investigator who is listed on the study and has completed all relevant training must be present.

AI cannot make medical decisions without human intervention, until such time as the AI tool has demonstrated superior decision-making ability.

Be sure to properly cite the tools that you use. You can talk to your instructor about the best way to do this, or if your instructor has no preference for citation format, you can check with UVM’s Library Citation Guide for current advice.

In any assignment, students must disclose their use of generative AI tools in completion of assignments including reflections and case reports, specifying tools used. Failure to do so is against the Code of Academic Integrity and will be responded to in accordance with the Enforcement of University and Larner Behavioral Standards Policy.

Use of AI in graded activities without permission and/or citation is against the Code of Academic Integrity and will be responded to in accordance with the Enforcement of University and Larner Behavioral Standards Policy.

Faculty must disclose their use of generative AI tools when developing educational materials and student assignments, fostering a culture of transparency and setting an example for students by openly sharing information about the data used for training, the challenges encountered, and the evaluation methods employed.

Faculty must disclose if AI products are used for analysis of student performance (e.g. OSCE or note review and feedback).

The Center for Teaching and Learning does not recommend using AI detection tools. Currently available tools are unreliable. False positive rates can be high. False negatives are also common.

Use of AI in graded activities without permission and/or citation is against the Code of Academic Integrity and will be responded to in accordance with the Enforcement of University and Larner Behavioral Standards Policy.

In any assignment, students must disclose their use of generative AI tools in completion of assignments including reflections and case reports, specifying tools used. Failure to do so is against the Code of Academic Integrity and will be responded to in accordance with the Enforcement of University and Larner Behavioral Standards Policy.

Each of us has an obligation to clearly communicate to students about how and why our stance on AI and other technology has shaped the course, its content, its assignment and/or its policies.

AI tools offer many possibilities for instructors and students. Course and assignment goals are the foundation of all decisions about AI. Your approach to incorporation or exclusion of AI will likely vary course to course and assignment by assignment; it’s best to think about what fits your particular learning goals. Transparency and clarity with students are the best ways to keep the emphasis on learning with integrity.

Faculty must disclose their use of generative AI tools when developing educational materials and student assignments, fostering a culture of transparency and setting an example for students by openly sharing information about the data used for training, the challenges encountered, and the evaluation methods employed.

Faculty must disclose if AI products are used for analysis of student performance (e.g. OSCE or note review and feedback). Faculty are ultimately responsible for ensuring the accuracy of the results generated.

These AI applications have been approved for use at UVM, and we either run the service ourselves or have contracts with the vendor that ensures UVM data is secure, private, and remains owned by UVM.

Individual tools are approved for use with data at various risk levels – Low, Moderate, High, or Restricted – as defined in the UVM Data Classification Matrix

When a service is listed as approved for a risk level, that means it is not approved for higher risk levels. For example, if a service is approved for “Low to High Risk”, that means it is not approved for Restricted data such as Protected Health Information or Controlled Unclassified Information.

All other general AI tools (ChatGPT, Claude, Gemini, etc.) Public information only

Any use cases other than Public Information require Contract Review

Microsoft Copilot Low to High Risk allowed Available to faculty, staff, and students

Allowed data: Low to High Risk, when logged in with your UVM NetID (Enterprise Data Protection)

Inappropriate use: uploading student health information; uploading UVM non-public data while using Copilot without NetID login.

High-risk data such as UVMHN protected health information (PHI), personally identifiable information (PII), non-public protected data (NPPD), or sensitive research data can never be entered into public domain AI solutions, such as Chat GPT, Bard, Bing, etc.

The University of Vermont has a contract in place with Microsoft Copilot but does not have a contract or agreement with any other AI tools or services. This means that standard UVM security, privacy, and compliance provisions are not in place when using technologies, other than Copilot.

If researchers want to use an AI tool with data determined to be moderate or high-risk/restricted, the AI tool is subject to applicable university policies around contracted purchasing, including an ISO Review.

The following folks are helping lead broad AI efforts across the university, including efforts around AI governance, strategy, workforce development, and more. The university has over 40 folks from across the university engaged in the AI taskforce to ensure a university-wide and thoughtful approach.

Notice: This page will be continually updated as AI policies and guidelines for researchers and research staff are adopted.