Why AI is Not Like a Calculator

When I was in third grade, I found myself with my first real academic struggle. I could not, to save my life, memorize my multiplication tables.

This was incredibly frustrating for me. Besides my terrible handwriting, I had never received a bad mark in a classroom. Yet, here I was, unable to memorize these tables no matter how hard I tried.

I did what many eight-year-olds do in that situation: I cheated, I used a calculator. Like most eight-year-olds, I was terrible at it and was caught immediately. The teacher was understanding but firm. She explained that I needed to do this because, in the future, I wouldn’t always have a calculator with me.

Obviously, she failed to predict the rise of cell phones. But I can’t fault her. It was 1988.

As an adult, I’ve come to recognize that everyone, including myself, has academic limitations. My brain is simply not wired for rote memorization. It is better at extracting meaning and interpreting things than memorizing them.

I never did learn my multiplication tables. Though I got better at them, there are still obvious gaps. However, it never impacted my life. Between cell phones and computers, I’ve never had to rely on my memory to multiply two numbers.

That is likely true for many people, which is why some compare calculators to generative AI systems. However, this has several problems, especially regarding AI in the classroom.

AI is not a calculator. And thinking of it like that is actively dangerous.

How Calculators Helped

To be clear, cheap, accessible calculators have been a boon for mathematics education. They’ve enabled students to solve more complicated problems with greater accuracy. Students and teachers are less bogged down by solving basic equations and can move on to more complex math skills.

However, that doesn’t mean students are excused from understanding how those basics work. There are still places in education where calculators are (and should be) forbidden or limited. Once students understand one skill, using the calculator to automate it speeds the progress to the next.

First, understanding how and why something works is important. Even though I ended up carrying a calculator with me everywhere, I still needed to know what multiplication is, including how it works, when to do it, and why.

Perhaps the push for rote memorization was unwise. But that was only a small part of that lesson. The focus was on multiplication itself. That was a lesson I did learn and have used regularly.

Second, though calculators don’t (or at least shouldn’t) make mistakes, humans using them do. If you punch 6×9 and get 36, you should recognize that you mistyped. Understanding the underlying process makes students better users of calculators.

Technology in the classroom is always a partnership between the student and the tool. Students who don’t understand what the tool is doing or how it works can not use it effectively.

And that brings us to the problem with AI.

The Problem with AI

Simply put, generative AI is not a calculator. It is infinitely more complicated, which opens up many new possibilities and creates many new dangers.

The first and most obvious difference is that calculators don’t make mistakes. If you punch in 6×9 and get something other than 54, you were the one who made the error. AI systems, on the other hand, are known to hallucinate. They state untrue things as fact, often without clear reason.

AI systems also have biases. They are developed by imperfect humans and trained on content by other imperfect humans, and their output often reflects the worst of that.

Humans can spot these issues when they are looking for them. However, that’s only if they understand the underlying process. If a human has never been taught how to write without bias or perform actual research, they will struggle to catch these errors.

One way calculators and AI systems are alike is that both can, in some situations, eliminate work. However, neither is a substitute for knowledge. The difference is that if I don’t understand multiplication, I can still get the correct answer on a calculator. If I don’t understand research, I may get an incorrect answer from an AI.

Trusting AI systems without understanding their goal is actively dangerous. This is why many programmers frown upon “vibe coding.” If you don’t know how to program, you can’t understand the AI’s output, or whether it’s safe, efficient, and accurate.

The problem is that the knowledge needed to correct or understand AI can take a lifetime to acquire. There are no shortcuts to understanding. Using AI without that understanding puts you at the mercy of its mistakes.

AI isn’t a calculator. Humans shouldn’t treat it like one.

Moving Forward

While this may sound like an anti-AI screed (and I’m sure some will take it as such), it’s not. AI is a very powerful and interesting tool. It’s also very new. We are still figuring out when and how we should integrate it into our lives, our work and our schools.

As I said in my AI usage gradient, there are many uses between “Entirely Human Written” and “Entirely AI Generated.” The focus shouldn’t be on dismissing AI or letting it replace human knowledge, but finding a balance where it can help with the work while keeping humans in control.

For one, there are uses of AI where understanding is less important. For example, brainstorming ideas for an article or essay is likely a good use of AI. It’s unimportant to know where or how the AI came up with those ideas.

Another example would be teaching students how to create an outline for an essay. Then, have the students generate, proof, and grade AI-generated outlines. Once students understand what makes a good outline, they can use AI to create outlines because they can dismiss the bad ones.

The problem with AI is that it’s a tool that can often be wrong. However, humans will never know if they don’t understand the problem they’re asking AI to solve. However, far too many people are trying to use AI as a substitute for understanding.

That is outright dangerous and has no place in the classroom.

Bottom Line

Integrating AI into the classroom will mean that there are spaces where it is not allowed. Students need to build their knowledge to use AI effectively. In that regard, it is very much like a calculator.

However, when a calculator makes a mistake, it’s because the human using it did. AI is often wrong, regardless of the human input it received.

That, in turn, is my most significant problem with AI. Every time I’ve dabbled with it, it’s never been much of a time saver. Validating and editing the work it generated has always been about the same or more time-consuming than just doing it myself.

AI has helped me with proofreading/editing human-created works and helped me generate ideas. I’m still open to finding areas where AI can assist, but those uses have to be both ethical and accurate.

AI struggles with the latter, and the humans using it struggle with the former.

Still, I agree that AI needs to be part of classroom education. However, its use cannot be a substitute for student knowledge. AI is not a calculator. It cannot be thought of as one.

It needs to be introduced and taught with the weight and gravity that comes with this technology. Comparing AI systems to calculators, if anything, is demeaning to AI.

AI needs to be seen for what it is, a powerful tool that requires knowledgeable humans to use and oversee it.