Teaching Isn’t Obsolete, but Our Assignments Might Be

I recently received an email from a community college professor in California. They’d heard I advocate for behaviorism as the foundation for teaching and learning. Their question was simple and sharp:

"If we define learning as behavior, what’s the point? Once you identify the expected behavior, ChatGPT is going to replicate it. So what can faculty do that ChatGPT can’t?"

It’s a fair challenge and it gets to the heart of why behaviorism is more relevant than ever.

Behaviorism Isn’t Just About Output. It’s About Conditions

Let’s be clear: ChatGPT can generate outputs. It can produce essays, simulate conversations, and summarize research articles. All of these are examples of observable behaviors. As B.F. Skinner, the founder of radical behaviorism, emphasized, the focus of learning must remain on the behaviors themselves and the environmental conditions that shape them. From this perspective, the question isn’t how these outputs were generated or what tool was used, but rather: who is performing the behavior, under what conditions, and how does that behavior change over time? While machines can simulate behavior and even adapt based on feedback, they are not users of technology; humans are. In education, our focus is not whether AI systems evolve or improve, but whether students learn. Students interact with tools like ChatGPT in ways that reflect growth, reflection, and context-sensitive performance because they are the ones engaged in a learning process that includes feedback, adaptation, and skill development. That’s what makes learning observable and meaningful, and that’s where instructional design comes in.

Behaviorism doesn’t diminish the value of the outputs we expect from students, it asks us to ensure those outputs reflect learning. The key question is: Did the student’s own behavior change? Did they act more skillfully, make more informed decisions, or respond more effectively than before?

Whether a student uses a pen, a keyboard, or a chatbot, what matters is whether they are the one performing the skill and whether our environment reinforces that behavior.

And this is why behaviorism is so powerful, especially in online education.

Faculty can’t always observe what students are doing, especially in asynchronous classes. But we can still observe what students submit, how they respond to feedback, and whether their behavior changes over time. That’s all the data we need.

Reproduction Is a Skill, but It Has to Be Demonstrated

In a chemistry class, students may need to replicate a chemical equation. That’s not rote, it’s precision. It’s a skill. If a student practices and performs it reliably, we are observing learned behavior.

But what if the student uses ChatGPT to generate the answer and pastes it into Canvas without further interaction? They’re still learning, just not what we intended. They’re learning that the environment rewards completion, not competence.

That’s not a student failure. That’s a design issue.

Every student’s behavior is an opportunity for learning. Behaviorism doesn’t blame the student. It asks: What did we reinforce? And if we don’t like the response, we change the task. We adjust the environment.

AI Isn’t the Problem. It’s Part of the Environment

Behaviorism doesn’t oppose tools. It asks:

Is the tool helping shape new behavior or replacing the need for it?

Students using ChatGPT to rephrase ideas, check logic, or refine their arguments can be learning deeply. These actions align with the kinds of observable behaviors described in Bloom’s Taxonomy, such as applying, analyzing, and evaluating. But learning only occurs if students are engaging with the tool in ways that change their own performance over time.

The problem is not that ChatGPT can generate answers. The problem is when our assignments don’t ask for anything more.

If the task is so predictable that a chatbot can complete it without human adaptation, then we are not observing student learning. We are observing task delegation.

And again: that’s not the student’s fault. It’s ours.

Every student's response is valid. It’s observable. It’s data. If we want different outcomes, we need different conditions.

Behaviorism Empowers Faculty in the Age of AI

Behaviorism doesn’t assume we can examine what is inside the student’s mind. It assumes we can observe what they do and use that data to adjust instruction and improve outcomes. That’s not a limitation; it’s a call to design environments that produce meaningful action.

Education Essential Reads

Fighting the Passive Learning Trap

Can We Just “Read” Audiobooks Instead?

STEM scientists, astronomers, physicists, chemists, and biologists, for example, enthusiastically embrace the newest technologies to advance their fields. They don’t question whether a telescope or a particle accelerator undermines the integrity of discovery. They ask: What can this tool help us do better?

So why should ChatGPT be treated as suspect in education? It is a tool. And just like any scientific instrument, it can either clarify or distort, depending on how it's used.

Behaviorism doesn’t dismiss the use of AI, it embraces it, as long as it helps shape student behavior aligned with the skills we want to observe: analyzing, evaluating, and creating. These are the verbs of Bloom’s Taxonomy. They’re not just standards, they are descriptions of behavior. And they’re meaningful when students perform them, not the tool.

In asynchronous classes, we might not watch students in real time, but we can:

• Ask them to narrate their decision-making process
• Design tasks that require them to explain a change or choice
• Provide feedback that changes how they respond next time
• Assign unfamiliar scenarios to test transfer, not repetition

Behaviorism gives us this clarity: If the behavior changes, learning has occurred. If it doesn’t, we don’t blame the student, we revise the environment.

Final Thought: Behavior Over Belief

The power of behaviorism in education, especially now, is that it centers observable change. It helps us focus on performance, not perception, and competence, not compliance.

In a world where AI can generate content, we don’t need to ask who wrote it. We need to ask:

Who learned from it?

And that’s something students show us through action.