Teacher resource

AI Detectors, Academic Integrity, and Better Assessment

A Learning Intelligence approach for teachers

Artificial intelligence has made academic integrity more complicated. Teachers may suspect that AI was used improperly, but AI detector scores are not reliable enough to serve as proof of misconduct.

A Learning Intelligence approach focuses on evidence, professional judgment, student understanding, and assessment design, not simply on catching students.

The 30-second guide

Start here when AI use raises a concern.

Detector scores are signals

They are not proof and should never determine an academic-integrity decision on their own.

Look at the learning process

Review drafts and notes, then ask the student to explain their ideas, choices, and use of tools.

Use evidence and judgment

Compare multiple sources of evidence, follow policy, and keep the response fair and focused on learning.

Clarify→ Review→ Conference→ Compare→ Respond

Why AI detector scores are not proof

AI detectors estimate whether writing resembles text produced by artificial intelligence. They do not determine who created the work, how it was created, or whether AI use violated classroom expectations.

Inconsistent accuracy

Results vary between tools and become less reliable when writing has been edited, paraphrased, translated, or created through a combination of human and AI input.

False results

Detectors can flag human writing as AI-generated and fail to identify work produced largely by AI.

Lack of context

A detector does not know the student, their previous work, their learning needs, or the conditions under which the assignment was completed.

Equity concerns

Multilingual learners and students who translate or revise their own writing may be falsely flagged more often.

No verifiable evidence

Unlike plagiarism software, detectors do not identify a source or provide material that can be independently examined.

Easy avoidance

Students who deliberately misuse AI can often alter the text enough to avoid detection, while honest students may still be flagged.

False accusations can damage trust, create conflict with families, and undermine confidence in the assessment process.

When you suspect inappropriate AI use

Move from suspicion to a careful review of expectations, process, conversation, evidence, and policy.

Clarify the expectations

Review what students were told before the assignment began. Was AI allowed, limited to specific tasks, prohibited, or required to be cited or disclosed?

A concern should be connected to a clearly communicated expectation.

Review the learning process

Look beyond the final product. Examine:

Planning documents
Drafts and revisions
Notes and research
Version histories
Sources and citations
Prompts or AI disclosure statements
In-class work

The goal is to determine whether the documented process reasonably supports the final submission.

Conference with the student

Ask the student to explain:

The main ideas in the work
How the work was developed
Why particular choices were made
What sources or tools were used
What was revised and why
What they learned through the process

Approach the conversation as an opportunity to understand the work, not simply to confirm a suspicion.

Consider multiple forms of evidence

Compare the submission with:

Previous assignments
Classroom discussions
In-class or handwritten work
Oral explanations
Demonstrated skills
The student’s documented process

No single piece of evidence should determine the outcome.

Respond fairly and consistently

When concerns remain:

Apply school and board policies
Document the evidence considered
Avoid relying on detector scores alone
Give the student an opportunity to respond
Use professional judgment
Follow established academic integrity procedures

The response should be proportionate, evidence-based, and focused on learning.

Preventing AI misuse through better assessment

The strongest response to AI misuse is not better detection. It is stronger assessment design.

Teachers can make student thinking more visible through assessment practices that capture learning as it happens, not only the finished product.

Make the process visible

In-class writing and production
Drafts and process portfolios
Oral conferences and presentations
Performance tasks
Reflection and metacognition
Version histories
Source verification
AI disclosure
Documentation of prompts, decisions, and revisions
Opportunities to defend or explain the work

These approaches make it easier to recognize authentic learning while also allowing students to use emerging technologies responsibly.

The Learning Intelligence approach

The Learning Intelligence Framework (LIF) helps students learn, use, and innovate with technology while developing judgment, creativity, responsibility, and AI intuition.

Students are expected to

Question information
Document their thinking
Make purposeful decisions
Verify results
Revise their work
Explain their use of technology
Take responsibility for what they submit

This shifts the focus from whether a student used AI to how it was used, what decisions the student made, and what learning the student can demonstrate.

Bottom line

Respond with care, evidence, and consistency.

Exercise professional judgment. Use AI detector scores cautiously, never as proof. When concerns arise, examine the process, speak with the student, compare multiple forms of evidence, and follow established policy.

The most effective long-term response is to design assessment that makes thinking visible and teaches students how to use artificial intelligence critically, transparently, and responsibly.