Integrating AI On Campus: Balancing Policy, Innovation, and Impact
Putting AI into action often begins not with algorithms, but with leaders who understand the importance of aligning emerging technology with real organizational needs. At Fairleigh Dickinson University (FDU), AI’s journey from idea to implementation has been guided by a deliberate focus on stakeholders and strategy. Helping drive this initiative is Devendra Mehta, Digital Strategy and Data Analytics Officer at FDU. With experience in data governance, cloud transformation, enterprise systems, and digital innovation across multiple industries, Mehta’s career has centered around aligning technology strategy with organizational goals.
Before joining higher education, Mehta built a foundation in using technology to drive measurable outcomes across various organizations and industries, driving success in their endeavors. That approach carried over to FDU, where he was initially brought on to support the rollout of IT Service Management (ITSM). The emphasis wasn’t just on tools, but on people, process, knowledge and organizational change resulting in a successful launch and a positive experience for the entire university community. As FDU’s digital needs evolved, so did Mehta’s role. What began as an ITSM initiative organically recently expanded into something larger: the conception, development, testing, and now rollout of the University’s AI program.
“I see a future where AI is woven into the University’s core operations. In the near term, faculty will be using AI to further enhance teaching and personalized feedback, while administrative staff will be automating high-volume tasks to focus on higher-value work. Students will likely benefit from AI-powered academic and career support, and research can accelerate through AI-assisted data analysis.”
– Devendra Mehta
Digital Strategy and Data Analytics Officer,
Fairleigh Dickinson University
Prioritizing AI Initiatives on Campus
“As FDU builds its AI capabilities, our approach to decision-making will be guided by three core lenses: governance, enablement, and value creation, while still leaving room for experimentation. On the governance side, we continue providing guidance for the safe, responsible, and compliant use of AI through strong data safeguards, policy frameworks, and risk reviews. From an enablement perspective, we’re helping academic and administrative units with their use cases so current and new technology, coupled with AI can be certain to streamline operations, enhance learning experiences, and improve decision-making. Finally, on value creation, we are reviewing all request to ensure initiatives deliver measurable outcomes, particularly those that expect to increase efficiency and support faculty innovation. Across all of this, we are intentionally creating space for quick wins along with experimentation, learning, and discovery, recognizing that AI is a rapidly evolving area with significant potential for new opportunities and breakthroughs.”
“Everyone brings a different perspective to AI, shaped by their own experiences and ideas, including viewpoints that we, as technologists, might not naturally consider. What feels familiar to some can be entirely new to others, and that diversity of thinking is something we want to enable. We are expecting our AI to allow us to accelerate data compilation and generate insights much faster, which opens the door to new questions and ways of thinking. Not every idea has to be driven solely by ROI. Through our governance model, we expect to create space and determine collaboratively which ideas are worth exploring from an experimentation standpoint, recognizing that innovation often begins before the value is fully defined.”
This past October at EdgeCon Autumn, Mehta joined Jeffrey Rubin, Senior Vice President for Digital Transformation and Chief Digital Officer, Syracuse University, to lead the panel discussion, AI in Action, Real World Applications and Outcomes of the New Higher Education Paradigm. “Universities are under growing pressure to implement AI, but adoption varies widely,” says Mehta. “Some institutions are still evaluating platforms, while others are already deploying AI agents and realizing tangible benefits. We were invited to share what was, at the time, our upcoming implementation, an initiative that has since gone live and was developed through a partnership between FDU and the University of California San Diego.
“Together, we were able to represent the full spectrum of AI maturity in higher education, from institutions early in the learning curve to those with more advanced deployments. It was a valuable opportunity to show that most universities fall somewhere along that spectrum and are working to implement AI in ways that are scalable, secure, reliable, and within their budget.”
“As FDU builds its AI capabilities, our approach to decision-making will be guided by three core lenses: governance, enablement, and value creation, while still leaving room for experimentation. On the governance side, we continue to provide guidance for the safe, responsible, and compliant use of AI through strong data safeguards, policy frameworks, and risk reviews. From an enablement perspective, we’re helping academic and administrative units with their use cases so current and new technology, coupled with AI can be certain to streamline operations, enhance learning experiences, and improve decision-making. Finally, on value creation, we are reviewing all requests to ensure initiatives deliver measurable outcomes, particularly those that expect to increase efficiency and support faculty innovation. Across all of this, we are intentionally creating space for quick wins along with experimentation, learning, and discovery, recognizing that AI is a rapidly evolving area with significant potential for new opportunities and breakthroughs.”
– Devendra Mehta
Digital Strategy and Data Analytics Officer,
Fairleigh Dickinson University
“This will mean starting with pilot-first approaches, focusing on targeted areas with measurable results, and collaborating with faculty and staff who are closest to the work and best positioned to shape meaningful use cases. Governance will continue playing a vital role. While it needs to remain iterative as policies evolve, institutions can build on the data security, privacy, and governance frameworks they already have as a strong foundation.”
As FDU rolls out an internal LLM-GPT model, the University is taking a pilot-focused approach to ensure adoption is practical, impactful, and guided by campus expertise. “Our rollout continues to be deliberate, working with early adopters and campus champions who are already passionate about AI and its potential to improve workflows or explore new pedagogical approaches,” explains Mehta. “We’ve engaged with faculty innovators, our AI committee, and peers from the academic and administrative sides. The most success we’ve had are with those who are open to discussing their current processes to identify flows and data points, and are committed to providing feedback on the pilot’s success. Beyond that, we ask them to serve as champions and help socialize the initiative, insights, and uncover new areas where AI can add value within their departments and with their peers.
“We began piloting our LLM-GPT model in late summer 2025, with the expectation of a rollout in the early fall, and officially launched toward the end of October. The pilot involved two teams: an alpha team, a small cross-section of academics and administrators, and a larger beta team spanning both areas. Their feedback helped us identify use cases to add to our roadmap. The official launch took place during an IT town hall, which also allowed us to reach the broader campus community, making it a very valuable experience. With the pilot now live and available to faculty and staff, the next phase focuses on finalizing our governance model and managing requests for additional AI agents. Each request will undergo discussion and technical review to determine the best way to implement it, as some may fall under generative AI while others may not.”
Defining Data Ownership and Governance
When using LLMs, institutions must carefully determine which campus data can be shared, which must remain protected, and how to enforce those boundaries effectively. “We follow our data security and privacy policies to ensure that regulated and sensitive information is reviewed and approved before placing it into the LLM” says Mehta. “Implementing our own LLM has provided us with a secure platform with role-based access for strong protections. Enforcement relies on technical controls, policy oversight, monitoring, auditing, and careful contract management with third-party providers to ensure our data is handled responsibly.”
“When working with third-party AI vendors, we ensure all contracts go through a process where we review model training, data transparency, and how our data may be used. We assess their security posture, data retention policies, protections in place. While these practices align with our broader data security policies, the speed and complexity of AI require additional rules, such as clear and immediate notification if our data will be exposed to AI or shared with third parties.”
As AI adoption accelerates on campuses, Mehta says establishing strong governance and policies should come first. “Governance frameworks around data for AI is essential, because without the most accurate and up to date information, AI becomes ineffective, prone to hallucinations, and potentially dangerous, while becoming a source of frustration for the users. Once a user loses trust with the system, it becomes difficult to regain it. Campuses need clear definitions of data ownership, quality standards, access controls, and a reasonable use policy for AI. Privacy, bias, workflows, and acceptable use all require careful attention, because AI can amplify existing issues rapidly. Policy requirements can help also guide decisions around infrastructure design, platform selection, and for general AI program development.”
Looking ahead, FDU envisions AI becoming an integral part of campus operations, enhancing teaching, research, and administrative workflows while creating new opportunities for innovation and learning. “I see a future where AI is woven into the University’s core operations,” shares Mehta. “In the near term, faculty will be using AI to further enhance teaching and personalized feedback, while administrative staff will be automating high-volume tasks to focus on higher-value work. Students will likely benefit from AI-powered academic and career support, and research can accelerate through AI-assisted data analysis.
“Our internal LLM, Fred GPT, the FDU Responsive Educational Database and named by a student, is positioned to be a trusted, campus-wide platform, securely supporting faculty, staff, and administration. The key will be getting AI into everyone’s hands and trained on how to effectively and properly use this continuously evolving tool. Access will allow for experimentation, learning, and gradually building more ROI-driven, impactful solutions that can expand an institution’s capabilities and competitiveness in an evolving educational landscape.”