Assessment in the Age of AI: Who’s Worried?

Wherever educators gather right now—staff meetings, conferences, professional learning sessions—the conversation about artificial intelligence is rarely far away. And almost without exception, a second conversation follows closely behind: What does AI mean for assessment and evaluation?

These conversations are often framed around concern. How do we prevent students from using AI inappropriately? How do we confirm and give consequences when they have? How do we “outsmart” a rapidly evolving technology? These are understandable questions if they reflect a sincere desire to protect the integrity of learning. Conversations about task design and the thoughtful use of AI are necessary. However, if these questions serve only to maintain the status quo, they risk limiting meaningful instructional growth.

But there is another group of educators who are noticeably calmer in the face of AI. They are not worried about assessment and evaluation, not because they are unaware of the challenges, but because they understand something fundamental: the architecture of sound assessment and evaluation has not changed in the light of AI.

The teachers who are not worried see assessment and evaluation as deeply human work. They understand that learning is a complex process, and that educators are insight-rich professionals who live alongside learners every day. These teachers do not outsource judgment to tools, policies, or single tasks. They know their students. They notice patterns. They listen carefully. They pay attention.

They understand that evidence of student learning must be centred on triangulation—the intentional collection of evidence over time from multiple sources: products, conversations, and observations. This collection of evidence happens in the classroom, alongside the learners.

Triangulation is not a new idea. It existed long before AI entered our classrooms, but the  external pressure AI exerts now compels teachers to more seriously reconsider established practices; it provides an opportunity—and a prompt—for that reflection.

The teachers who are not worried do not view assessment as a series of isolated events—tests, quizzes, projects, and assignments that happen at fixed points in time. Instead, they understand that everything a student says, does, or creates is potential evidence of learning.

A comment during a class discussion, a question asked while working through a problem, the way a student approaches a task, revises a draft, or explains their thinking to a peer—all of these moments matter. Learning unfolds in real time, and educators who are paying attention are constantly gathering evidence, often without needing to label it as such.

When assessment is understood this way, AI generated products are far less threatening. A single artifact has never told the whole story of learning, and it still does not. In these cases, an AI-generated product within that body of evidence adds to the story, rather than takes away.

The teachers who are not worried understand that professional judgment matters. For them, collecting evidence is not primarily about placing students in order or determining winners and losers. It is about uncovering the truth of what a student knows, understands, and is able to do in relation to the learning demands placed upon them. For one student, strong evidence may come primarily from rich conversations and repeated observations. For another, products may play a larger role. The mix is not fixed, and nor should it be.

This flexibility does not undermine fairness and equity. In fact, it increases them, because it allows educators to respond to learners rather than force learners to fit a predetermined assessment structure.

The teachers who are not worried do not confuse reliability and validity. They understand that the most important question in assessment is not “Is this easy to score?” but “Does this actually measure what it is intended to measure validly?” and “Are my findings reliable because they are repeatable across multiple sources over time?”

Depending on the outcome, concept, or competency, the appropriate evidence may vary. Consider a senior years chemistry outcome that requires students to perform a lab demonstrating the formation of solutions based on polar and non-polar substances. A written lab report alone cannot capture the complexity of this learning. The outcome demands observation: how students handle materials, follow procedures, respond to unexpected results, and apply safety protocols. In this case, observation is not optional—it is the only valid way to collect evidence aligned with the learning expectation. Teachers must allow evidence to come from the source that best aligns with the outcome, rather than defaulting to written products simply because they are easy to collect.

However, let’s push this a bit further. In the past, evidence collected from a product may, in fact, have been valid—that is, a strong match between the intended outcome and the evidence gathered. Increasingly, however, teachers may need to ask themselves, “How can students show what they know, understand, and can do in ways that privilege conversations and observations?” Because now, conversations and observations not only add richness to the body of evidence, but they also mitigate the worry that the written product has been outsourced to AI.

The teachers who are not worried are not asking questions such as, “What percentage of the final grade must come from products, observations, and conversations?” They know that these kinds of formulas often distract from the real work. They intentionally separate evaluation and reporting until just the right time.

Instead, they ask themselves: “What do I need to be certain that students know, understand, and can do what the learning demands of them?” They dig deeply into curricular documents rather than rely primarily on commercially prepared resources. That question leads naturally to thoughtful task design, ongoing evidence collection, and professional judgment—regardless of whether AI is present.

The teachers who are not worried do not wait until the end of learning to ask whether a product was created using AI. They are collecting evidence along the way. They know how ideas develop. They hear students explain their thinking. They see misconceptions surface and resolve. Because of this, academic dishonesty is rarely a surprise. When something does not align, they notice—not because of software, but because the learning evidence no longer makes sense when viewed as a whole.

I am reminded of my time working in New Zealand, when headlines emerged about university students paying overseas ghostwriting companies to write papers. This was framed as a shocking discovery, yet it was only shocking in systems where final grades depended heavily on a small number of high-stakes products and an exam. Limited tasks and high-stakes assessments have long been a source of anxiety for learners—but they should also invite concern and reflection from teachers, as they reveal how vulnerable our assessment and evaluation designs can be.

Educators who included conversations and observations, and collected evidence along the way would have known long before the headlines appeared. The disconnect between what students could explain, do, and apply—and what appeared on paper—would have been visible.

The teachers who are not worried are also open about their own use of AI. They reveal to students how AI supports their planning, thinking, and problem-solving. They teach students to use AI transparently, ethically, and intentionally, rather than pretending it does not exist.

This openness reduces fear and secrecy, and it reinforces that learning—not deception—is the goal.

In short, AI does not require us to abandon sound assessment practices. It challenges us to return to them. The teachers who are not worried understand that assessment grounded in triangulation, professional judgment, and deep knowledge of learners has always been resilient. In the time of AI, it remains so—not because it resists technology, but because it centres learning as a deeply human endeavour.

And that, ultimately, is what assessment has always been about.

Meet the Expert(s)

Sandra Herbst

Superintendent/CEO, River East Transcona School Division

Sandra Herbst serves as Superintendent/CEO of River East Transcona School Division in Winnipeg, Manitoba, where where her work is grounded in the belief that every person deserves to be wrapped in care, dignity, and love.

Read More