Generative AI is not just a new tool — it is a catalyst for reimagining how we teach, research, and serve our community. At Mississippi State University, we stand at the forefront of this transformation, committed to shaping the future of AI education and practice in our state and beyond.
To lead in this rapidly evolving landscape, supervisors are the cornerstone of our success. Importantly, you do not need to be an AI expert. Your greatest impact comes from being the catalyst — sparking curiosity, encouraging experimentation, and fostering a culture where learning and innovation thrive at every level. Success will not come from simply rolling out new software or offering routine training. Instead, it requires supervisors to champion change, empower their teams, and model openness to new ideas.
Supervisors set the tone for adoption and growth. By exploring AI tools, asking questions, and sharing experiences, you inspire others to embrace change. Your role is to guide, support, and celebrate the journey — ensuring every team member feels empowered to contribute, learn, and innovate, while knowing they have your encouragement and backing.
The mindset shift
AI requires a mindset shift from how we typically adopt new technologies. Rolling out a new software package and offering training for the few who will use it will not work. AI touches everyone and changes too fast. This mindset shift requires:
- Supporting innovation at every level, not just in central offices or for just a few employees.
- Fostering an environment where we help each other learn. Departments should provide in-house training and share successes with one another to accelerate adoption.
- Supporting in-house expertise and encouraging eager and talented individuals to design solutions for tasks that may or may not be part of their job.
- Fostering an environment where innovation becomes part of everyone’s role, not a separate activity or for just a few.
- Requiring skill development and recognizing that upskilling will be continuous as Generative AI becomes a natural part of daily work.
Everyone must assume some ownership in their own upskilling. Supervisors must lead. Supervisors should model thoughtful adoption, set clear expectations, and recognize innovation and impact.
Key actions for supervisors
1) Lead the way and designate a champion
- Lead by example: Supervisors should explore approved AI tools themselves so they can speak from experience when guiding staff and department use.
- Designate an AI Champion: Appoint an eager staff member willing to share best practices and serve as a point of contact for AI-related questions within the department.
- Departmental working group: Form a unit-level group to review practices quarterly, curate use-cases, share impacts, and adapt guidance.
- Maintain accountability: Supervisors are responsible for ensuring staff adhere to guardrails. Maintain a simple AI tool log to track what tools are used, what data were used, and for what purpose. Know what’s happening in your department.
2) Understand approved platforms and data protection
MSU has approved these tools:
- Microsoft Copilot (MSU tenant)
- ChatGPT Teams/Enterprise (when institutionally licensed)
- Webex AI Assistant
- Microsoft Teams
All other tools should be considered public or open and not suitable for MSU data designated as High-Risk or Regulated. These include:
- Gemini, NotebookLM, Grok, Claude
- Cloud environments at AWS, Nvidia, Google CoLab, or others (unless in an MSU instance of these platforms)
Do not use READ.AI for recording and transcription of meetings. Use MS Teams or Webex for this purpose. If you have used Read.AI and can’t clear it from your computer, submit a ticket to ITS for help.
Data classifications
It is important for you to know your data and understand the level of confidentiality required. Read AI Data Security at MSU for more details. It is impossible to anticipate every possible use. If you are uncertain, contact the Chief Information Security Officer at security@msstate.edu.
- High-Risk / Regulated (FERPA, HIPAA, GLBA): Request approval before using, and only use an MSU-approved tool. Example: Uploading student grades to draft a letter is not allowed without approval. While we know our MSU Copilot instance is secure, we prefer to know who is considering using highly regulated data in any AI platform.
- Moderate-Risk Data: Any data that requires logging into the MSU system with your NetID to obtain. Use only MSU enterprise-licensed tools like Copilot. Example: Using Copilot in Outlook (MSU tenant) to summarize emails or meeting notes that include internal project discussions. Using Copilot to analyze internal financial data or draft an employee performance report.
- Low-Risk / Public Data: Most other data. May be used in public tools. The recommendation is to change settings to not allow general LLM training on the data. Examples: Using Gemini to draft a public event announcement or graphic that will still be reviewed and approved by staff. Using NotebookLM to summarize a set of journal articles.
3) Human review before public release
Non-negotiable rule: AI-generated content must never be automatically emailed, posted online, or otherwise made public without human review.
Human review and approval are mandatory for:
- Emails to students, faculty, or external stakeholders
- Social media posts, website updates, or newsletters
- Visual content such as images, charts, or infographics
Not allowed examples:
- An AI agent sending automated reminders to students without human approval
- An AI agent publishing content directly to a department social media
Website AI chatbots: All public-facing AI chat assistants should be approved before development and again before deployment.
4) Quality, integrity, and bias
- Always verify facts, references, and citations in AI outputs. Example: confirm that policy citations are correct and current.
- Check for bias in language, especially in HR, student-facing, or public communications. Example: ensure marketing materials avoid unintended exclusionary language.
- If asked to disclose whether AI assisted you, always be truthful.
- Version control: AI drafts are working notes only. The human-reviewed, approved document is the official record. A best practice: store all AI-generated drafts in one folder and move final, human-reviewed versions to another folder for sharing or storing.
5) Performance reviews
- Be transparent about how AI use factors into evaluations.
- Employees remain responsible for the quality and integrity of their outputs, whether AI assisted.
- Consider when you want to require disclosure of AI-assisted work. Example: a footnote in a report stating, “Portions of this report were generated with AI assistance.”
- Set appropriate goals for learning to use the tools depending on roles and responsibilities.
6) Research and compliance
- NIH is the only agency with specific AI guidance — see Supporting Fairness and Originality in NIH Research Applications (NOT-OD-25-132).
- Restricted research data such as export- or ITAR-controlled must follow very specific guidelines. Discuss with the Office of Research Security.
- PIs should carefully read all solicitations and contracts for agency requirements. Direct any questions to the Office of Research Compliance and Security.
7) Training and capacity building
- Promote training: Encourage staff to attend MSU-offered workshops, online training, and vendor-led modules.
- Encourage using the LinkedIn Learning platform, available to all MSU employees and students.
- Peer learning: Host short demos and “show and tell” sessions where staff share practical use cases. Consider a departmental Team where employees can share AI practices.
- Supervisor role: Ensure all staff are aware of training opportunities and encourage attendance.
8) Recognize innovation and promote continuous improvement
- Recognize innovation: Celebrate AI use that saves time or improves outcomes. Recognize this during annual reviews. Example: a staff member who uses AI to streamline meeting agendas and minutes.
- Continuous improvement: Hold quarterly unit-level sessions to discuss both successes and risks encountered in AI use.
For supervisors of hands-on and operational teams
Many teams at MSU — facility maintenance, landscaping, custodial services, and other hands-on roles — may not immediately see a direct connection to AI adoption in their daily work. Supervisors in these areas should know there is no expectation for rapid or widespread adoption of AI tools where they do not naturally fit. All teams play a vital role in keeping our campus running smoothly, and the value one brings is not diminished by the pace of AI technology change.
While you don’t need to worry about integrating AI into your team’s work, over time, new technologies may emerge to support repairs, scheduling, safety, or other operational tasks. If you have ideas or when use-cases arise, let us know — we’ll provide guidance and support to help you and your teams explore them.
Encourage your teams to participate in all university learning opportunities. We don’t want to leave anyone out who wants to learn.
Supervisor checklist
- Explore and learn.
- Identify an AI Champion and form a working group.
- Inventory data and tasks for potential AI use; flag moderate/high-risk data.
- Verify tools against the approved list; escalate exceptions to VP for approval.
- Publish unit-specific AI norms.
- Require human review before all public-facing AI content.
- Implement quality checks and confirm sources.
- Document examples of positive impact.
- Hold quarterly “AI lessons learned” sessions.