Rethinking Assessment When Your Students Have Access to ChatGPT

---

A Grade 10 English teacher assigns a five-paragraph essay on the themes in To Kill a Mockingbird. Half the class submits eerily polished work. The other half submits writing that clearly reflects their own voice.

This scenario is playing out in Canadian classrooms every day. Traditional assessment methods were not designed for a world where students have access to tools that can generate competent written work in seconds.

The question is no longer whether students are using AI. The question is what we do about it.

Why AI Detection Is Not the Answer

Many teachers have turned to AI detection tools like GPTZero and Turnitin’s AI detection feature.

But these tools have limitations. They generate false positives. They are particularly unreliable for English language learners. And they operate in an arms race they cannot win — as AI models evolve, detection struggles to keep pace.

More importantly, reliance on detection creates an adversarial classroom dynamic. Education works best when built on trust, not surveillance.

The Opportunity Behind the Problem

AI can produce products. It cannot demonstrate process.

Most provincial curricula emphasize higher-order thinking: analysis, creativity, collaboration, reflection. Yet many assessment tasks still measure only final output.

Assessment in the AI era must shift from product to process.

Seven Assessment Strategies That Work in 2026

1. Process-Based Assessment

Require brainstorming notes, research logs, drafts, and reflections. Assess the learning journey — not just the final submission.

2. In-Class Composition with AI as a Tool

Allow AI use — but require students to document prompts, outputs, revisions, and reflection on their decisions.

3. Oral Assessment and Conferencing

A five-minute conversation often reveals deeper understanding than a written essay alone.

4. Application to Local Contexts

Ask students to apply learning to their own community, lived experience, or novel classroom-specific scenarios.

5. Collaborative Assessment

Design tasks where knowledge is built collectively and individual contributions are documented.

6. Authentic Performance Tasks

Shift from describing knowledge to demonstrating it through projects, experiments, debates, and real-world applications.

7. Transparent AI Classification Per Assignment

Label assignments clearly: “No AI,” “AI as brainstorming,” “AI as partner,” or “AI-focused.” Transparency reduces confusion and builds trust.

What This Means for Equity

Not all students have equal AI access at home. If AI is expected, ensure equitable access through school resources.

At the same time, be cautious about penalizing AI use in ways that disproportionately impact English language learners or students with disabilities.

Assessment in the AI era must prioritize fairness as much as integrity.

The Bigger Picture

If AI can produce a competent five-paragraph essay in 30 seconds, perhaps that essay was never the strongest measure of learning.

AI has forced a long-overdue question: Are we assessing what truly matters?

Critical thinking. Authentic communication. Application. Reflection. Collaboration.

AI did not create this question. But it has made it impossible to ignore.

---