“I know I need to address this behavior, but I don’t know how to start the conversation.”
“I need to communicate a complex message clearly, but I’m not sure how it will land.”
“How do I give this feedback without damaging the trust I’ve worked so hard to build?”
For many university managers, these moments are becoming the starting point for a new kind of collaboration, one with generative AI. Across the University, supervisors are already using tools like Gemini to draft communication and organize their work. In a recent Supervisory Conversations session on AI in supervisory work, many participants shared that they are already experimenting with these tools, with the majority of people using it to support communication. Some described using AI to rephrase emails for clarity and tone, especially when navigating sensitive situations or communicating across different audiences. As adoption grows, so does the need to be intentional about how AI fits into supervisory practice.
Where AI Helps, and Where Managers Lead
AI operates at a speed that can feel almost effortless. It can summarize meeting notes, generate talking points, or help draft a difficult message in seconds. In that sense, it is a powerful support tool, especially for the administrative and preparatory parts of supervision.
But speed is not the same as judgment.
AI cannot create psychological safety in a tense conversation. It cannot read the room, notice what is NOT being said, or take responsibility for a decision that affects someone’s work or experience. These are not gaps that will be “fixed” with better tools. They are fundamentally human responsibilities. As managers, our role is shifting away from being information holders and toward being relational anchors, people who bring clarity, accountability, and steadiness when it matters most.
This is where AI can be most useful, not as a replacement, but as a way to prepare.
Supervisors are already using AI to role-play difficult conversations, draft initial feedback language, or organize complex information before a meeting. For example, you might ask Gemini to act as an employee who is resistant or defensive, allowing you to practice how you would respond while staying calm and focused. Others shared using AI to outline key points before a one-on-one or to translate a vague concern into more specific, behavior-based feedback. These uses can reduce cognitive load and help you enter conversations more prepared. The key is recognizing that the output is a starting point, not a final answer.
It might be less technical than you think
As conversations about AI evolve, so does the idea of “AI literacy.” It is no longer just about knowing how to write a good prompt or understanding how probability models work behind the scenes to predict the next word in a sentence. Instead, it is about developing the skills to work with AI intentionally. Research from Anthropic highlights several behaviors that distinguish more effective use, which can be summarized as follows:
- Iterate. Treat the first response as a draft. Ask follow-up questions, refine the request, and stay in the conversation until the output truly meets your needs.
- Question polished outputs. When something looks complete or well-written, that is often the moment to pause. Ask yourself: Is this accurate? Is anything missing? Does this reflect my context? This is especially important when drafting communication, where tone, nuance, and context matter.
- Set the terms of the collaboration. Many users never tell AI how they want it to respond. Being explicit, for example, asking it to push back on your assumptions, explain its reasoning, or surface uncertainty, can significantly improve the quality of the interaction.
The common thread across all of these behaviors is not technical skill. It is judgment.
AI can help you think, draft, and prepare. It cannot decide, connect, or lead. You remain accountable for how information is interpreted, how decisions are made, and how people are supported. Several supervisors in our January session emphasized this point directly, noting that while AI helped them prepare, the responsibility for the conversation and its impact still sat fully with them.
The opportunity is not to become dependent on AI, but to become more intentional in how you use it. Iterate. Question. Set expectations. Continue to refine your own expertise so you can evaluate what AI produces.
AI can support your work, but you remain responsible for the decisions you make and how you show up as a leader.
