Having Productive AI Conversations

Investigation, Not Accusation

When AI detection flags student work—or you suspect AI use—the response matters as much as the finding. Accusatory approaches damage relationships and sometimes punish innocent students. Investigative approaches protect everyone.

The Accusation Approach (DON'T):

"The detector says you used AI. This is cheating. You'll receive a zero."

The Investigation Approach (DO):

"I noticed some patterns in your work I'd like to understand better. Can we talk about your writing process?"

The difference is profound: one assumes guilt, the other seeks understanding.

The Investigation Framework

Step 1: Observe, Don't Conclude Detection flags are data points, not verdicts. Gather more information:

• Review the detection report details
• Compare to student's previous work
• Look at revision history (if available)
• Check for unusual patterns or style shifts
• Consider student's context (ESL, learning style, etc.)

Step 2: Invite Conversation Approach privately and without accusation:

"Hi [Student], I'd like to talk about your recent essay. Can you meet with me briefly after class/during lunch/[specific time]?"

Don't say why yet—this prevents defensive preparation.

Step 3: Ask Open Questions Start with curiosity, not interrogation:

• "Walk me through how you approached this assignment."
• "What was your process for writing this?"
• "What resources did you use?"
• "What part was most challenging?"
• "What are you most proud of in this work?"

Listen more than you speak. Let them tell their story.

Step 4: Go Deeper if Needed If answers seem incomplete, probe further:

• "Tell me more about [specific section]."
• "This paragraph sounds different from your usual style. What happened here?"
• "I notice the draft history is limited. Can you tell me about your revision process?"

Step 5: Consider All Evidence Before concluding:

• Does their explanation make sense?
• Does it match other evidence (drafts, in-class work, oral explanation)?
• Could there be innocent explanations?
• What's their history with integrity?

Conversation Scripts

Script 1: Initial Inquiry

"Hi [Student]. I wanted to talk about your [assignment] for a few minutes. There are some patterns in the writing I'm curious about. Can you walk me through your process for creating this?"

Wait for response. Take notes. Stay neutral in expression.

"That's helpful. Can you tell me more about [specific element]? What sources or tools did you use?"

Based on response, continue investigation or move to next script.

Script 2: When AI Use Seems Likely

"I appreciate you sharing your process. I want to be direct with you: some elements of this work look like they may have been generated by AI tools like ChatGPT. My detection tool flagged some patterns, and [specific observation, e.g., 'the style differs from your in-class writing'].

I'm not here to trap you—I'm here to understand what happened and help you learn. Can you tell me more about what happened?"

Give space for response. Avoid interrupting.

"I hear you. Here's what I'm thinking about..."

Script 3: When Student Admits AI Use

"Thank you for being honest with me. That takes courage. Let's talk about what happens next.

First, help me understand: What was going on that led you to use AI this way? Were you overwhelmed? Confused about the policy? Running out of time?"

Understanding why matters for addressing root causes.

"Here's how we'll handle this: [explain consequences per your policy]. But more importantly, I want to make sure you get the learning from this assignment. Here's what I'm thinking..."

Script 4: When Student Denies AI Use

"I hear that you're saying this is your own work. I want to believe you, and I'm going to give you a fair process.

Here's what I need from you: [specific evidence request—drafts, outline, verbal explanation of content, etc.]. This will help me understand your work better.

If you have evidence of your process, please bring it to me by [time]. If you don't, we'll need to have another conversation about how to move forward."

Script 5: When Uncertain

"Honestly, I'm not sure what happened here. The detection tools aren't perfect, and there could be explanations I'm not seeing.

Here's what I'm going to do: For this assignment, I'm going to [specific action—perhaps reduced grade with opportunity to redo, or oral defense]. Going forward, I'd like to see [specific process requirement].

I'm giving you the benefit of the doubt this time. But I also want you to know that I'm paying attention. Fair?"

Focus on Learning, Not Punishment

The Real Goal: When students use AI inappropriately, the goal isn't punishment—it's ensuring they actually learn what the assignment was meant to teach.

Reframing Questions: Instead of "How do I punish this?", ask:

• "What learning did this student miss?"
• "How can I help them genuinely learn this material?"
• "What led them to make this choice?"
• "How can I prevent this situation in the future?"

Learning-Focused Consequences:

Punitive Approach	Learning Approach
Zero on assignment	Redo assignment (unassisted, supervised)
Automatic failing grade	Oral exam on same content
Disciplinary referral	Conference about skill gaps + support plan
Public humiliation	Private conversation + learning contract

The Redo Option: Many integrity violations are better addressed by having students demonstrate their learning through alternative means:

• In-class rewrite of same essay
• Oral presentation of their argument
• Timed exam on the same content
• Annotated revision showing original thinking

This ensures the learning happens while still having consequences.

Root Cause Analysis

Understanding why students use AI inappropriately helps prevent future incidents:

Common Root Causes:

Root Cause	Signs	Response
Time pressure	Multiple late assignments, stressed demeanor	Time management support, check workload
Skill gaps	Struggled with similar assignments before	Targeted skill instruction, scaffolding
Unclear expectations	Confusion about policy, didn't know it was wrong	Clearer communication, explicit policies
Disengagement	Pattern of minimal effort, doesn't see relevance	Connection to interests, relevance discussion
Outside pressure	Mentions parents, grades, college	Holistic support, perspective discussion
Opportunity	Assignment design made AI too easy/tempting	Redesign future assignments

Root Cause Questions:

• "What was going on when you worked on this assignment?"
• "What would have helped you complete this on your own?"
• "Were there obstacles I should know about?"
• "Looking back, what would you do differently?"

Documentation Best Practices

Why Document:

• Protects you if accusations are disputed
• Creates record for patterns
• Supports fair, consistent handling
• Helps administration if escalation needed

What to Document:

Element	Example
Date and time of discovery	"Noticed patterns on 10/15, 3:00 PM"
What triggered concern	"Turnitin flagged 68%, style shift from draft"
Student conversation summary	"Met 10/16. Student explained process as..."
Evidence gathered	"Reviewed Google Doc history, shows minimal editing"
Student response	"Student acknowledged using AI for two paragraphs"
Action taken	"Assigned oral defense of essay, will redo if fails"
Follow-up plan	"Check-in scheduled for 10/23"

Template:

AI Integrity Investigation - [Student Name] - [Date]

Assignment: [Name]
Trigger: [What raised concern]

Conversation Date: [Date]
Conversation Summary:
[Brief notes on what was discussed]

Student Explanation:
[What student said]

Evidence Reviewed:
- [List evidence]

Conclusion:
[What you determined]

Action Taken:
[Consequences/response]

Follow-Up:
[Next steps]

When to Escalate

Not every situation requires escalation, but some do:

Handle Yourself:

• First-time offenses
• Clear learning opportunity
• Student cooperative
• Policy was unclear
• Minor impact

Consider Escalation:

• Repeated offenses
• Student defiant or dishonest in investigation
• High-stakes assessment
• Pattern across multiple classes (if known)
• Parent involvement likely

How to Escalate:

1. Document thoroughly first
2. Know your school's process
3. Present facts, not conclusions
4. Include student's explanation
5. Recommend appropriate response
6. Be prepared for various outcomes

Protecting Yourself and Students

Protect Students By:

• Never publicly accusing
• Using investigation, not assumption
• Considering all explanations
• Applying consequences fairly
• Focusing on learning
• Documenting process

Protect Yourself By:

• Following school policy
• Documenting everything
• Avoiding conclusions without evidence
• Involving appropriate parties when needed
• Being consistent
• Staying calm and professional

Key Takeaways

1. Use investigation, not accusation—approach with curiosity, not judgment
2. Conversation scripts help navigate difficult discussions professionally
3. Listen more than you speak—let students explain before concluding
4. Focus on learning, not punishment—the goal is ensuring students learn
5. Understand root causes to prevent future issues and address real needs
6. Document everything thoroughly for protection and consistency
7. Know when to escalate and follow proper procedures when needed

:::