Generative Artificial Intelligence guidelines
July 8, 2024
Best practices for using new Gen AI tools
As Generative Artificial Intelligence (Gen AI) tools become more common in the workplace, we wanted to share some responsible use guidelines created by UVA Information Technology Services.
- What is Gen AI? A computer or computer system that is capable of generating new content (such as images or text) in response to a submitted prompt (such as a query), for example Copilot and ChatGPT which can be used to generate written text or documents or DALL-E or Adobe Creative Cloud features which can be used to generate images.
These tools can be used for enhancing productivity, learning and creativity but they also pose some risks and challenges. UVA ITS has developed some responsible use guidelines (listed below) for using these tools at UVA.
AI Guidelines
- Understand liability issues — UVAs licensed Generative AI tools provide data protection that does not exist when using individually licensed tools.
- Use UVA-licensed AI tools whenever possible — UVA's licensed Generative AI tools (UVA Copilot Chat and UVA Copilot for M365) include contractual data protection of university information.
- Adhere to all existing University policies, including:
- IRM-002 — Acceptable Use of the University's Information Technology Resources
- IRM-004 — Information Security of University Technology Resources
- IRM-012 — Privacy and Confidentiality of University Information
- IRM-017 — Records Management
- HRM-009 — Preventing and Addressing Discrimination and Harassment
- HRM-041 — Policy on Sexual and Gender-Based Harassment and Other Forms of Interpersonal Violence
- Know that not all answers are accurate or appropriate — All AI generated content should be reviewed by a knowledgeable person before using or publishing.
- Consider the appropriateness of using AI generated content — If you, or the University, would suffer reputational harm if it became known that you used AI generated content for an assignment, task, official communication, etc. then you shouldn't use AI.
- Be mindful of data bias potential — AI models are trained on large datasets that may contain biases present in the data.
- Be transparent — Consider informing individuals when AI-enabled tools are being used and/or when content has been AI generated.
- Be knowledgeable in prompt engineering — Be thoughtful of how you construct your prompts to help produce relevant responses.
- When using UVA-licensed Generative AI tools, you agree to a specific set of Terms of Use, which include the following guidelines and additional items not listed here:
- AI Users should only use GenAI Tools in circumstances when GenAI Tools may enhance or assist in performing academic or job-related tasks, such as by enhancing productivity, learning, efficiency, and decision-making (not for commercial or personal purposes).
- No AI User may use GenAI Tools for personnel decision-making purposes without the express written consent of the Vice President and Chief Human Resources Officer (or an appropriately authorized designee).
- GenAI Tools should not be used blindly for decision making and/or the creation of content without the review of a knowledgeable person, and should never be relied upon for important inquiries. AI Users are expected to recognize the limitations of the GenAI Tools they are using, avoid over-reliance on such tools, carefully review output for errors, and remain vigilant to identify potentially erroneous, incomplete, or otherwise problematic output.
If you have questions or would like to learn more about using Generative AI at UVA, contact the FM IT Help Desk at (434) 982-5579 or fmhdticket@virginia.edu.
FM Technology & Innovation in partnership with Occupational Training will host an upcoming Lunch & Learn session on this topic set for 12-1 p.m. Wednesday, Aug. 21 offered virtually and in-person in Skipwith Hall Room 112.
Jane Centofante
Communications senior generalist
UVA Facilities Management
(434) 982-5846
janecentofante@virginia.edu