Ethical use is critical when applying AI tools to an academic setting.
These guidelines, developed by the UCSB ITC Subcommittee on AI, are intended to serve as guidance for members of the campus community who engage with AI for research, teaching, administrative work, and other university-associated functions. They are intended to be applied in contextually appropriate ways to a rapidly evolving set of AI conditions.
We encourage you to review the University of California's AI Principles to learn more about ethical considerations before working with AI.
Accuracy, Reliability, and Safety
AI-enabled tools should be effective, accurate, and reliable for the intended use and verifiably safe and secure throughout their lifetime.
Privacy and Security
AI-enabled tools should be designed in ways that maximize privacy and security of persons and personal data.
Fairness and Non-Discrimination
AI-enabled tools should be assessed for bias and discrimination. Procedures should be identified, depending on the concrete tool and deployment context, and put in place to proactively identify, mitigate, and remedy these harms.
Shared Benefit and Prosperity
AI-enabled tools should be inclusive and promote equitable benefits (e.g., social, economic, environmental) for all. Knowledge of available AI-enabled tools and access / instruction / training to AI-enabled tools created and used in a university context should be made broadly available to the community.
Appropriateness
The potential benefits and risks of AI and the needs and priorities of those affected should be carefully evaluated to determine whether AI should be used, or if the use of AI should be endorsed, discouraged, or prohibited. Before individuals or units make decisions about AI, potential benefits and risks should be carefully evaluated, along with the needs and priorities associated with the context for AI usage.
Human Values
AI-enabled tools should be developed and used in ways that support the ideals of human values, such as human agency and dignity, and respect for civil and human rights. Individuals or units implementing AI tools should ensure that adherence to civil rights laws and human rights principles is examined as part of AI tool adoption where rights could be violated.
Transparency
Individuals should be informed when AI-enabled tools are being used. When individuals are permitted or forbidden to use AI tools, or when individual or campus unit data is used to train AI-enabled tools, this should be made clear by the units implementing the AI tools. The methods being used to gather the data and provide them to the AI tool and the purpose(s) for doing so should be explained. Individuals should be able to request that their data not be used, and remedies to address harms should be in place should they occur.
Accountability
The University of California Santa Barbara should adopt appropriate policies, processes, and structures to ensure that the University consistently enacts and pursues adherence and accountability to the above principles in its development, use, and regulation of the use of AI systems. In implementing AI tools the University and/or its relevant units should provide a clear process for individuals to express concerns about their use. The University should remain aware of general AI developments, especially in conjunction with enterprise-level technologies for which the University has contracted, and ensure that these are consistent with the principles outlined in this document.
AI Technical & Security Guidelines
These guidelines focus on deploying artificial intelligence (AI) solutions while upholding data security, privacy, and ethical use to ensure the responsible and secure adoption of AI technologies at UC Santa Barbara. These guidelines should be used by those units and individuals responsible for the technical implementation of AI solutions at UCSB.
1. Data Management and Protection
Privacy:
- Ensure third-party AI systems comply with privacy regulations such as FERPA and GLBA.
- Collect, store and process only the necessary data to minimize privacy risks.
- Minimize or mask personal information to reduce the risk of re-identification.
Security:
- Verify vendors and/or solutions use strong encryption for data at rest and in transit.
- Vendors must also delete all UC data when a contract ends.
- Data Proximity: Prioritize architectures locating data near AI compute capacity for cost and performance efficiency.
2. Vendor Evaluation
Security Reviews:
- Assess the inclusion of AI third-party applications, including privacy policies, certifications, and audit reports.
- Consult with the Office of Information Security to assist with reviewing the use case and AI implementation. Users who seek to incorporate P3/P4 data should contact the Chief Information Security Officer's office and may also need to contact UCSB Human Resources for employee data. They may also need to contact the Campus Privacy Officer.
- All AI implementations must follow UC information security policies, including required security reviews.
Transparency:
- Request documentation on how AI models process data and generate outputs.
Data Usage:
- For third-party tools, choose AI engines that do not use UCSB prompts or data for training. UCSB-developed tools and/or models must utilize and store UCSB data in a manner consistent with both these guidelines and the UCSB Responsible AI Principles.
3. Deployment Best Practices
Piloting:
- Test AI technologies in controlled environments before full deployment.
Training:
- Training data must be anonymized
Integration:
- Ensure solutions integrate securely with UCSB infrastructure using compliant APIs.
Continuous Improvement:
- Be ready to replace AI engines to take advantage of improved performance, reductions in cost, or vendor misbehavior.
These guidelines are intended to assist those seeking to implement artificial intelligence (AI) tools or services at the University of California, Santa Barbara. AI tools considered for implementation at UCSB should be evaluated and prioritized according to at least the following general criteria:
Implementations whose outcomes are well-aligned with core elements of the campus mission should be prioritized above those that are not. In particular, implementations that align with the articulated goals and objectives of the campus IT strategy should be preferred over those that do not.
Implementations that provide greater benefit to the university and its academic and administrative activities relative to their cost should be given priority over those that provide less. Value should be measured in terms of improved experience for campus stakeholders, new/enhanced capabilities, increased efficiency and/or effectiveness in operations, and other commonly recognized benefits to the university and its community members. Value projections based on current successful use cases are superior to those that are largely hypothetical. Cost should be measured both by expenditures for hardware, software licenses, and services, as well as by time/labor expended by UCSB staff, faculty, or students.
Implementations that have lower risk should be given priority over those that have higher risk. Risk should be measured in terms of the complexity of implementation, likelihood of overall success, and the probability of occurrence of negative outcomes. A useful catalog of risk considerations for AI implementations is provided by the UC AI Council Risk Assessment Guide.
Innovation is a core value for UC Santa Barbara and the campus’ distributed operational structure supports this by delegating decision-making across campus divisions and departments. To support effective utilization of resources in this environment UCSB has established IT governance processes and bodies to oversee major technology implementations.
Given the near-universal applicability of the technology, AI implementations may arise for consideration in numerous parts of the campus. As a general guideline, those implementations with expected initial implementation costs of over $100,000, and/or ongoing operating costs of over $50,000, should be brought to the IT Council for review and evaluation. For more information on the IT Council and how to bring implementations for consideration, please contact Elise Meyer, ITS Director of Strategy & Academic/Research Support, at emm@ucsb.edu.
- Josh Bright - Associate Vice Chancellor for Information Technology and CIO, ITC AI Subcommittee Chair
- Ambuj Singh - Professor, Computer Science
- Ann-Marie Musto - Associate Vice Chancellor and Chief Human Resources Officer
- Ben Price - Associate CIO, Administrative Business Systems, Service Management & Automation
- Elise Meyer - Director, IT Strategy & Academic/Research Support
- Fabian Offert - Assistant Professor, History & Theory of the Digital Humanities
- Hector Villicana - Executive Director, Letters & Science IT, Associate CIO Instructional Computing
- Jackson Muhirwe - Chief Information Security Officer (CISO)
- Jeremy Douglass - Assistant Professor, English
- Joaquin Becerra - Dean of Students, Office of Dean of Students
- Joe Sabado - Deputy CIO
- Kelly Caylor - Professor and Associate Vice Chancellor for Research
- Linda Adler-Kassner - Professor and Associate Vice Chancellor, Teaching and Learning
- Miles Ashlock - Chief of Staff, Executive Director, Planning & Administration,Office of the Vice Chancellor for Student Affairs
- Robert Hamm - Assistant Dean, Graduate Division
- Ruimeng Hu - Assistant Professor, Mathematics
- Shea Lovan - Chief Technology Officer
Considerations for Campus Populations
When using AI in the context of your role on campus, you can set yourself up for success by asking these questions
- Does generative AI tool enhance teaching and learning experiences while maintaining academic integrity?
- Am I being transparent about the use of AI-generated content in course materials?
- Am I encouraging students to use AI responsibly and ethically?
- Am I complying with University policies governing behavior when using AI in the academic environment?
- Am I using generative AI to improve administrative processes and decision-making?
- Am I ensuring data privacy and security when using AI tools in daily tasks?
- Am I staying informed about the latest AI developments and best practices?
- Am I complying with university policies governing workplace behavior when using AI?
- Am I using generative AI to support learning and research while adhering to academic integrity policies?
- Am I properly citing AI-generated content in academic work?
- Am I aware of the limitations of AI-generated content?
- Am I verifying its accuracy before using it in academic or research contexts?
- Am I complying with university policies governing behavior when using AI in the academic environment?