AI Did Not Break Assessments. It Exposed What We Were Actually Measuring.

One of the biggest concerns I keep seeing around AI in education is assessments.

Teachers are worried that:

• students are using AI to complete assignments
• students are turning in work that is not really theirs
• and traditional assignments are becoming harder to validate

I understand the concern.

But honestly, I do not think the problem is as complicated as people are making it.

The bigger issue is that AI is exposing something that already existed:

Many assignments were designed to evaluate the final product more than the actual learning process.

AI just made that weakness easier to see.

---

The Real Starting Point

The first thing teachers need to do is clearly define the actual goal of the assignment.

What are we really trying to measure?

Is it:

• Reading comprehension?
• Evidence of understanding?
• Application of knowledge?
• Writing ability?
• Critical thinking?
• Argumentation?
• Content mastery?

Because once the goal is clear, the assessment process becomes much easier to design.

---

The Rubric Matters More Than the Tool

The next step is building a rubric that clearly identifies:

• what evidence matters
• what proficiency looks like
• and how student understanding will actually be evaluated

That is important whether AI exists or not.

A lot of assessment problems happen because:

• the goal is vague
• the rubric is unclear
• or the assignment is measuring completion instead of understanding

AI did not create those issues.

It just made them more visible.

---

What AI Is Actually Good At

One of the mistakes people make is assuming AI is “thinking.”

It is not.

AI is fundamentally a probabilistic pattern recognition system.

That matters because pattern recognition is actually very useful in education when applied correctly.

Instead of using AI as the final evaluator, I use it to help identify patterns in student work and generate follow-up questions based on the student’s submission.

At that point, the process changes completely.

The question is no longer:

“Did the student use AI?”

The better question becomes:

“Can the student explain, apply, and defend their understanding?”

That is a much more meaningful assessment question.

---

Ownership Verification Quizzes (OVQs)

In the software I have been building, I call these models:

Ownership Verification Quizzes (OVQs)

Although when presenting them to students, I usually call them:

Application and Explanation Quizzes (AEQs)

The purpose is not to “catch” students.

The purpose is to shift assessment back toward:

• understanding
• application
• reasoning
• and explanation

The model identifies patterns in the assignment and then generates targeted questions connected directly to the student’s own work.

That creates a very different instructional conversation.

---

Why This Has Changed My Teaching

This approach has made my instruction much more meaningful and efficient.

Instead of spending large amounts of time trying to determine whether a student “cheated,” I can spend time discussing:

• their reasoning
• their understanding
• their choices
• and how they arrived at their conclusions

The assessment itself becomes another learning opportunity.

Sometimes the student understands the material very well.

Sometimes they reveal gaps in understanding almost immediately.

Either way, the conversation becomes far more valuable than simply grading a final product.

---

AI Did Not Remove the Need for Teachers

If anything, AI makes teacher judgment even more important.

The solution is not removing rigor.

The solution is designing assessments that:

• prioritize understanding
• require explanation
• connect to reasoning
• and keep humans involved in the evaluation process

That is also why I do not think the future of assessment is:

• AI grading everything automatically
• or trying to completely block AI usage

I think the future is building systems where:

• AI helps identify patterns
• teachers guide the process
• and students demonstrate ownership of their learning

---

Final Thought

I do not think AI broke assessment.

I think it exposed weaknesses in how we were already measuring learning.

The good news is that it also gives us an opportunity to build better systems that focus more on:

• understanding
• application
• explanation
• and meaningful instructional conversations

And honestly, I think that is probably a better direction for education anyway.