I'm in a master's program with a focus in AI...
But, during my experience, most instructors don’t even acknowledge AI use in class.
They may offer some sort of note about “integrity” or a warning with no examples. But, day to day, the message is pretty much silent.
Meanwhile, we all know what’s happening – students use ChatGPT (or Gemini, or Claude, or others) to outline, to debug, to ideate, to get feedback on assignments, to write for them, etc. It’s not a secret, but most people (students + teachers) don’t talk about it openly (i.e., shadow AI usage).
And that “not speakable” part is where the main issue lies.
When such a powerful tool is readily available but “lives” outside of the rules of the classroom, it shapes nervous systems instead of the learning. People may start managing risk instead of building the skill, or optimizing for speed and avoiding questions that would help them grow and use their critical thinking skills.
In fact, I see this same pattern in organizations. For example, a new hire uses AI to write an email, edits it just enough to sound “human,” but never discloses the use of AI to their manager. They are not trying to “cheat,” and this is not an AI problem. The issue lies on the lack of psychological safety.
It’s understandable and a safety response – not a character flaw.
Here’s how to look at it through a positive lens: this moment is not about the student’s inability to “think.” It is simply proof that we need a new norm that protects both learning and mental health. When we name what’s true, shame disappears, leaving space for judgment to improve!
The goal? Don’t police AI. Do teach how to use it confidently and transparently.
Make AI Use "Speakable," Then Make It Skillful
The Path of Least Resistance
- Start with a clear permission statement
- Add it to the syllabus and clearly discuss what counts as allowed use, what doesn’t, and why. Clarity is the best medicine for anxiety, in this case.
- Ask for an “AI use note” as reflection
- Just a quick section on what you used, what you changed or edited, what was iterated/rejected, and how you verified. This completely changes it to a learning mindset (oh, and cite your use of AI in the sources).
- Grade the process
- Reward clear evidence of thinking: source review, logic checks, bias spotting, and others. This is a difficult change as grading takes more time than just grading a final output. However, I’d argue that moving away from the “result” and assessing the “path” to the answer is the best way to measure actual human value in this new AI world.
- Teach “human-in-the-loop” as a wellbeing skill
- Verifying the work is not just “busywork, boring work.” It is how you stay grounded when a tool such as AI sounds so incredibly confident. Verification builds self-trust.
- Model it like a leader
- Be clear and transparent about your own AI use and workflows: where it helped, where it failed, and how to correct it. Transparency increases trust.
These are just some examples of how to strengthen the standards, along with sending a mental-health-friendly message that you don’t need to hide to succeed here.
Silence or pretending AI isn't here...
Doesn’t stop its use. Instead, it increases its “shadow use,” teaching people to do it alone.
Higher education can and must be the place where AI confidence and best practices are learned. Make AI normal to discuss, safe to practice, and rigorous to verify. With AI here to stay, this is how you can reduce “Shadow AI” risks and help current and future generations graduate trusting and confident about their own thinking.
And for those leaders hiring new graduates? The same applies – focus on what’s happening rather than looking away, and start building a culture where it’s always safe to be transparent.
