Suddenly, that same student struggles to explain their reasoning, construct a coherent argument, or recall content they seemed comfortable with during homework.

For many students, the issue is not necessarily dishonesty. Rather, they never fully engaged in the thinking process that develops true understanding and skill mastery. When AI completes the practice for them, they miss out on the learning that practice develops.

This scenario is becoming increasingly common as students offload portions of their learning to generative AI. In the first two blogs of this series, I argued that the real concern surrounding AI is not cheating, but rather, lost learning and that AI literacy and thoughtful classroom culture must be part of the solution. But what does that look like day to day?

AI literacy alone does not ensure students are actually doing the learning. That is why AI-resilient formative assessment is critical to protecting the learning process. More than ever, the process matters as much as the product, and the process deserves our attention.

Cognitive Offloading vs. Cognitive Augmentation

A useful framework for this discussion is the distinction between cognitive offloading and cognitive augmentation.

Cognitive offloading occurs when students rely on AI to do the core thinking or problem-solving for them rather than using it to support understanding. Signs that this may be happening include polished work students cannot explain, limited ability to describe their reasoning, and little evidence of drafting or revision (Schwartz and Diliberti).

Cognitive augmentation, by contrast, occurs when AI supports deeper thinking rather than replacing it. This may look like students brainstorming after generating their own ideas, critiquing AI responses, identifying limitations or errors, revising drafts, or using AI to practice and review content. In these situations, AI supports learning rather than eliminating the productive struggle that learning requires (Pearson: School Educator Guide).

As teachers redesign formative assessments, the question shifts from:

"Did students use AI?"

to:

"How can I design formative assessments that require students to do the thinking?"

Why Formative Assessment Matters More Now

Traditionally, formative assessment was used primarily to monitor progress and assess mastery.

Today, it serves another critical role: making student thinking visible.

As discussed in Blog 2, trust, transparency, and AI literacy are essential. But those values only matter if students have opportunities to actively demonstrate their thinking. When redesigning formative assessment for an AI-enabled classroom, a useful question becomes:

How do I make student thinking visible during the learning process?

Formative assessment gives teachers a window into how students are learning. It reveals misconceptions, creates opportunities for feedback and discussion, and allows students to practice responsible AI use while understanding is still developing.

What Visible Thinking Looks Like

For many teachers, making thinking visible begins by reconsidering their learning environment. Some are eliminating technology entirely for portions of learning and returning to paper-based work. This is one great way to ensure thinking remains visible.

When that is not practical or desirable, another approach is to shift expectations away from answers and toward reasoning.

Rather than only asking:

"What is the answer?"

ask:

"Why is this the right answer?"

Even better, ask students to explain their reasoning orally or through video reflections. Those forms of evidence are significantly more resistant to AI substitution because they require students to articulate their own thinking.

Evaluate Your Existing Formative Assessments

One helpful exercise is to run your own formative assessments through a commonly used large language model.

For example:

"I am a 10th-grade student who received the following assignment: 'Identify the main idea of Chapter 9 of The Great Gatsby and explain why in 4–5 sentences.'"

Most AI tools will provide a polished, grade-appropriate response almost instantly.

Knowing this, consider revising the task:

• Ask AI to generate three possible main ideas from the text.
• Select one.
• Support your choice using evidence from class discussion or the text itself.
• Reflect on how AI influenced your thinking.

This revision shifts the task from answer generation to analysis, judgment, and reflection.

Four Small Upgrades Any Teacher Can Try

1. Require Drafts and Checkpoints - Learning lives in drafts, feedback, and revisions. Consider how those moments can become visible through handwritten work, oral check-ins, revision histories, or staged submissions.
2. Ask Students to Explain Why or How - The quality of a student's explanation often reveals far more than a correct answer. Prioritize reasoning alongside accuracy.
3. Use Reflection Prompts or AI Logs - Ask students to reflect on specific class discussions, connect ideas to their own experiences, or document when and how they used AI during the learning process. Reflection and transparency support both learning and integrity.
4. Vary Assessment Formats - Mix oral responses, video explanations, observation notes, written reflections, collaborative discussions, and annotations. Different formats provide different windows into student understanding.

Looking Ahead

Remember, this can start small.

Choose one or two formative assessments in a unit and experiment. See what works. Continue the conversation with students and remember that many of them care deeply about developing their own critical thinking (Schwartz and Diliberti; Walton Family Foundation et al.).

Design for process. Let students into your thinking. Learning integrity will follow.

If formative assessment helps us make learning visible, a new challenge emerges:

What happens when it is time to evaluate learning?

How do we rethink projects, essays, presentations, and other product-based assessments when AI can contribute so heavily to the final result?

That is the focus of the next blog in this series.

Stay tuned.  

References:

Pearson. Assessment Evolved: Redefining Formative Assessment in a Generative AI Era. Pearson, 2026.

Pearson. School Educator Guide: Redefining Formative Assessment in a Generative AI Era. Pearson, 2026.

Schwartz, Heather L., and Melissa Kay Diliberti. More Students Use AI for Homework, and More Believe It Harms Critical Thinking. RAND Corporation, 2026.

Walton Family Foundation, GSV Ventures, and Gallup. Gen Z Is Using AI—But Reports Gaps in School and Workplace Support. Walton Family Foundation, 2025.