Good Practices in the Use of Generative AI

The use of AI tools, especially generative AI tools, is increasing dramatically in the day-to-day business of institutions of higher education. This document is meant to serve as guiding principles to member institutions in the responsible and ethical use of generative AI.

Since the early 2020s—and especially after the release of GPT-3.5 in 2022—generative artificial intelligence (AI) has shifted from novelty to utility, steadily integrating into how people across higher education brainstorm, write, and solve problems. The term generative AI reflects a set of technological systems capable of producing new content such as text, images, audio, and video that are designed to feel strikingly human. The core technology of generative AI is large language models (LLMs), which are deep neural networks trained on massive datasets to model the structure and meaning of human language. After extensive training to model language, follow instructions, use tools, respond safely and helpfully, and tailored to respond in ways preferred by humans, these models can generate essays, summarize readings, answer questions, translate languages, and even simulate interactive dialogue. While generative AI currently commands the spotlight, it builds on decades of earlier AI research, including machine learning, expert systems, symbolic reasoning, and knowledge representation.

Assistants and agents are two emerging formats that expand how generative AI is being used and deployed. Assistants are customizable systems built around LLMs, often shaped by "system messages" or "custom instructions" that define their behavior, tone, custom knowledge content, and scope. Users can tailor these assistants to take on roles such as writing coach, instructional designer, or policy consultant. Crucially, assistants can go beyond generating text. They can be connected to external tools like code interpreters for performing calculations or generating graphs, web search for retrieving current information, or private knowledge sources such as uploaded PDFs or documents. These enhancements allow assistants to ground their responses in accurate data, reducing hallucinations and expanding their usefulness for domain-specific tasks.

Agents build on these capabilities by operating with a degree of autonomy. Rather than generating text responses as a single output, agents pursue goals through iterative reasoning: they break down tasks, plan actions, take steps, evaluate outcomes, and refine their strategies over time. They may call Application Programming Interfaces (APIs), use external software, or maintain a memory of past steps in order to handle more complex or evolving problems. This goal-oriented, tool-augmented approach positions agents not just as robust helpers capable of managing multifaceted academic or operational workflows.

In the context of colleges and universities, these tools are being integrated into an expanding range of academic and operational settings. Faculty use assistants to draft syllabi, analyze literature, or design assessments. Students engage with generative tools to revise writing, simulate interviews, or explore new topics interactively. Administrators and staff are exploring agents that can manage calendars, respond to routine queries, or synthesize meeting outcomes. As these systems become more deeply embedded in the digital infrastructure of higher education, they bring both new capabilities and new responsibilities—raising questions about authorship, accountability, pedagogy, and the long-term role of human expertise in academic institutions. This good-practices document is meant to provide food for thought as colleges and universities wrestle with institution-wide AI policies, guidelines, and strategies.

Institutional leadership is important as colleges and universities consider implementing an AI strategy. Because AI tools could be used by faculty, staff, or students in their official work and because those people come from every corner of the institution, AI strategies are by nature institution-wide concerns. The institutional priority and focus could be set by a governing board or the President/Chancelor or by any cabinet member with oversight over aspects affected by new AI technologies. For example, the Chief Academic Officer (CAO) might oversee an institution-wide strategy in teaching AI. The institution’s Chief Research Officer (CRO) might seek to provide AI resources to enhance the research enterprise on a campus. The Chief Information Officer (CIO) might address the need for a safe digital space to use generative AI tools without the fear of sensitive data being incorporated into external large language models. Ideally, each of these leaders would work together to create an institution-wide AI strategy that is consistent with and representative of the institution’s mission and values.

Key stakeholders also include board members whether governing (e.g., trustees) or advisory (e.g., an industry advisory board). The inclusion of board members with experience in industry can help improve workforce development outcomes, alignment with other governing bodies, and more. Local government can be very supportive as well and, in some cases, have provided considerable financial support for college and university-led AI efforts. Alumni and other friends of the institution provide not only development opportunities, but also hands-on experience in the AI sector providing valuable insight into the kind of new talent that industry is looking for.

One challenge in any institution-wide endeavor is communication. One strategy that has worked elsewhere is to name a point person whose job it is to move among different working groups to keep lines of communication open. This does not need to be a cabinet level leader, but the point person should have access to the right teams and institutional leaders to stay informed and keep others informed. We have seen that it is helpful to engage your institution’s communications team for both internal and external communications about your AI strategy.

There are many units typically found on college campuses that should be engaged. These include, but are not limited to, enrollment management, student advisors, career services, student life, centers for teaching and learning, centers for the use of technology in teaching, professional development (lifelong learning), and others that support faculty, staff, or students in their study, work, and growth.

As generative AI becomes increasingly embedded in teaching, research, coursework, and administration, institutions may need to revisit their policies to ensure they remain relevant, fair, appropriate, and enforceable for campus populations. Generative AI introduces new capabilities and risks that may unexpectedly intersect with long-standing policies in ways that were not previously anticipated. Areas for review may include:

• Information Security and Data Governance: Policies should account for the security implications of hosting or interacting with AI models, including third-party integrations, data sharing, and the generation of executable code or scripts.
• Classroom Recording, Coursework Redistribution, and Intellectual Property: With AI tools capable of transcribing lectures and students commonly generating new material from instructor-developed coursework to support, enhance, and deepen their learning, institutions must consider intellectual property and ownership of these course materials. The diversity in AI platforms regarding privacy and the use of prompts and user-provided content - information which the user may or may not have intellectual property rights to provide - to train new models and develop AI techniques may require a revisit of who owns academic content, how it can be reused or redistributed, and the rights and protections afforded to instructors and students regarding course content and materials. Institutions may need to consider mechanisms for enforcement or new technological platforms that can equitably provide solutions balancing the needs of all stakeholders.
• Created Works and Intellectual Property: Intellectual property guidelines should clarify whether AI-assisted content is considered original work and to what extent it is acceptable.
• Academic Integrity and Honor Codes: Institutions should explicitly define when and how AI assistance is permissible in coursework, exams, and research, helping students and faculty distinguish between legitimate support and academic misconduct. Institutions should clarify the extent to which AI-assisted content counts as original work, and the extent to which it is acceptable.
• Faculty Evaluation and Promotion: Criteria for evaluating teaching, research, and service may need to evolve to reflect the downstream effects of AI-enabled scholarship during the tenure and promotion process and consideration of broad service activities related to the training and applications of AI outside of the specifics of the discipline.
• Student Privacy and FERPA Compliance: The use of AI to analyze or generate insights from student records and communications may require updated guidance to maintain compliance with federal privacy regulations and institutional data ethics.
• Accessibility and Accommodations: As AI tools become part of the learning environment and the increasing gap in cost and performance of these AI tools, institutions should ensure they support, rather than inadvertently undermine equitable access and inclusion for all students.

The governance of AI in higher education institutions requires a multifaceted approach that balances innovation with responsibility, security, and integrity.

Role of the CIO: The Chief Information Officer (CIO) plays a crucial role in determining which AI applications and platforms are permitted in institutional operations, ensuring that they align with institutional cybersecurity best practices and protect intellectual property, data privacy, and compliance with regulatory frameworks.

Role of the CAO: The Chief Academic Officer (CAO) has a significant role in guiding faculty on AI use in syllabi, specifying whether and how generative AI tools may be used for assignments, and addressing concerns about overreliance or ethical misuse.

Role of academic departments: Academic integrity is a critical issue in the use of AI, requiring each academic department to provide clear guidelines on how AI-generated content is monitored, restricted, or integrated into coursework to prevent plagiarism and uphold ethical standards. Similarly, research integrity policies must ensure that AI-assisted analysis and content generation meet rigorous scholarly expectations. Other considerations in the creation of institutional and departmental AI policies include the guidelines provided by specific AI platforms for platform usage, privacy, and security (such as OpenAI’s guidelines on prompt engineering). Additionally, guidelines developed by grant funding agencies, academic peer review venues (journals and conferences), scholarly publications, and professional associations will be valuable sources of information in guiding the creation of AI policies at various levels of granularity. A continuum of AI usage policies must be considered with discipline-specific contexts at institutions.

Data Governance: Data governance is a key concern, necessitating policies that safeguard sensitive student and faculty information while enabling responsible AI use in research and learning. Key factors to consider include the regulatory and compliance frameworks, accreditation requirements, research ethics and research integrity guidelines, and security and privacy constraints of AI platforms.

Each institution will develop unique policies based on its mission, academic culture, and technological infrastructure, which may lead to tensions—such as conflicts between institution-wide AI governance and individual faculty policies outlined in specific syllabi, or more fundamental issues such as determining the scope of what is considered as AI usage. Addressing these challenges requires continuous collaboration between administration, IT leadership, faculty, and students to ensure that AI enhances education while upholding institutional values. Since AI tools are evolving rapidly, any policies concerning AI use in at least three different areas (acceptable use, prohibited use, responsible use) must be regularly reviewed.

Higher education institutions hold a critical responsibility to ensure that the use of generative AI aligns with principles of human-centered AI, emphasizing privacy, transparency, accountability, and ethical inclusivity. Institutions must actively assess and mitigate algorithmic biases, implement transparent practices around AI deployment, and safeguard both student and faculty data privacy. In addressing academic integrity, institutions must clearly differentiate between appropriate and inappropriate uses of generative AI to prevent plagiarism, establishing guidelines that foster originality while recognizing the legitimate role AI can play as a creative and analytical aid. Pedagogy should be reimagined to reflect this evolving landscape—focusing on critical thinking, process-oriented learning, and assessments that value students' ability to effectively integrate AI tools in ethical and meaningful ways. By doing so, institutions not only uphold academic standards but also prepare students to engage with AI responsibly in their academic and professional lives.

AI in teaching and learning: Integrating AI education across an institution is essential to preparing students for a workforce and society increasingly shaped by artificial intelligence. Graduates need AI literacy not only to compete in today’s job market but also to navigate and critically engage with the digital world in which they live.

Discipline-specific AI competency development in courses: Because AI is used in vastly different ways across disciplines, academic departments must have the flexibility to determine how AI is incorporated into their curricula to best serve their students.

Faculty resources: Equipping faculty with resources to thoughtfully integrate AI is key, and institutions can leverage centers for teaching excellence, student advisors, and career resource centers to support both instructors and learners in understanding AI’s role in their respective fields. Additionally, AI itself can be a valuable tool in higher education, helping faculty design new courses, develop innovative learning exercises, and personalize instruction.

Interdisciplinary AI-enhanced collaboration opportunities: Ensuring that AI is broadly taught and thoughtfully applied across disciplines will empower students with the critical skills and ethical awareness necessary for success in their careers and as informed users of AI tools and platforms.

Academic institutions can use AI to provide personalized, data-driven support tools to help students succeed, while still emphasizing human mentorship, ethical AI use, and academic integrity. The following are good practices and AI-driven strategies to enhance student achievement and retention.

AI-Powered Tutoring for Academic Achievement: AI functions as a smart tutor by offering students personalized academic support tailored to their individual needs. Adaptive AI tutoring systems can adjust to each student’s learning pace, identifying strengths and weaknesses and modifying instruction accordingly. Natural language processing (NLP) tools can provide instant explanations and alternative problem-solving methods, encouraging active learning through interactive tasks and real-world applications. To extend support beyond traditional hours, institutions can deploy 24/7 AI-powered chatbots to answer questions on course material, assignments, and common inquiries. These chatbots can be designed to escalate complex issues to human faculty when necessary, and can support multimodal formats—including text, voice, and video—to accommodate diverse learning styles. Additionally, predictive AI can analyze student performance and suggest personalized learning paths, offer real-time feedback, and recommend targeted exercises to address specific academic challenges.

AI-Powered Academic Advisement for Retention: AI can play a crucial role in academic advisement by guiding students throughout their educational journeys. Personalized degree planning systems driven by AI can recommend courses based on individual strengths, interests, and career objectives. These systems can help students balance their workloads and suggest alternate pathways if they encounter difficulties in certain subjects. Furthermore, AI tools can assess academic performance in conjunction with labor market trends to propose relevant career options. Resume and portfolio analysis tools can align student accomplishments with employer expectations, while AI can also match students with appropriate mentorship opportunities.

AI-Driven Early Warning Systems for At-Risk Students: To improve student retention, institutions can use AI to proactively identify students who may be at risk of academic failure. Predictive analytics can monitor attendance, assignment submission patterns, participation levels, login behaviors, and even sentiment in discussion forums to detect signs of disengagement or struggle. When early warning indicators arise, AI systems can generate alerts and create personalized intervention plans, including tutoring, study groups, or faculty engagement. These systems can send automated nudges via chatbots, emails, or text messages to keep students on track. Additionally, virtual assistants can provide deadline reminders and connect students with campus resources. Importantly, AI can work in tandem with human advisors, offering a dashboard that highlights risk levels and enables targeted outreach, ensuring that human support is always available when needed.

AI-Driven Student Engagement and Retention Strategies: AI can significantly boost student engagement and motivation. Through gamification, AI can tailor learning experiences with challenges and rewards, fostering a sense of achievement. For community building, AI can recommend peer study groups based on similar academic difficulties or learning styles, creating supportive networks. Moreover, AI can be instrumental in monitoring student well-being, detecting signs of stress or burnout and offering recommendations such as counseling, mindfulness practices, or wellness resources. By combining personalization with proactive support, AI helps foster a more engaging and resilient academic environment.

Offices of institutional planning and effectiveness bear the responsibility for collecting, analyzing and reporting institutional data. The reports are often compiled for internal use, external stakeholders (students, parents, community), state, and federal agencies. The data volume created by even a small institution of higher education can be substantial and AI tools can be very beneficial in making sense of disparate data. Further, generative AI tools can be helpful in synthesizing material submitted to institutional planning offices, generating reports, and communicating results. However, institutional planning offices should continually evaluate and adapt their use of AI tools to meet best practices. Security and privacy of sensitive data should be a chief concern when applying AI tools in this context. Being certain of data privacy policies with AI tools and especially whether your data are unknowingly leaving your institution is of primary importance.

Similarly, offices of institutional planning should be aware of the policies around the use of generative AI that may come from agencies to which they are providing reports. Local, federal, and other agencies have sometimes banned the use of AI tools outright rather than delineate the specific ways in which tools can be used.

Institutions of higher education are currently facing many challenges; how to create a comprehensive AI strategy is one of them. This good-practices document is offered to provide high-level guidance to institution leaders. Because AI strategies impact the entire institution (faculty, staff, students, community members and stakeholders), an institution-wide approach is needed. Most higher education institutions are struggling with creating comprehensive AI strategies given the rapid nature of AI development. Some institutions have found it helpful to confer with similar higher education institutions to share ideas. Regardless of the details of any institution’s AI strategy, of primary importance is the ability to be nimble and pivot given the incredibly rapid pace of AI development.