The pitfalls and the promise of offloading our thinking

By Kim Taylor   •    December 10, 2025

Commentary/Op-Ed

This blog post was inspired by Kim’s presentation AI & Critical Thinking: Empowering the Next Generation at the 2025 Science Writers and Communicators of Canada conference.

Math Wars and the Calculator Debate

(Steve Buissinne via Pixabay)

Look at the picture in the above. Do you know what it is? Have you ever used one yourself? You might know that this is a handheld calculator. You might even own one of these.

This nifty little tool might seem quaint these days, but I remember the outcry from parents when they first began to appear in schools. These devices were part of the ‘math war’ that led many to question how calculators would impact math education.

Would students become overly reliant on these machines? Would they lose the ability to do basic math? Would they use the technology to ‘cheat’ their way through homework and tests? Even 50 years later, educators still come out on different sides of the calculator debate.

In June 2025, I asked participants at the 2025 Science Writers and Communicators of Canada (SWCC) conference during our "AI for Good" panel discussion if they considered using a calculator to be cheating. Most said no. But one person remarked that if a student used one during a test, then yes, it would be cheating.

Others quickly agreed, saying that the appropriateness of calculators depended on the situation. It could be inappropriate for an elementary student to use a calculator for a calculation, but not for an adult to use one in everyday life for the same calculation.

Using devices such as calculators means we can exert less mental effort, thus freeing our brains to do more complex tasks. Similar concerns are now being expressed about generative AI (gen AI) by parents and caregivers, educators and business leaders. They wonder how gen AI tools are impacting students’ thinking and are concerned that students will become overly reliant on these tools.

(Graphic by SWCC)

Gen AI and Cheating

Today, students have access to increasingly sophisticated gen AI tools, such as ChatGPT, that can help them write assignments, do homework and ace tests. And many students use them.

Given students’ easy access to gen AI, it’s not surprising that a 2025 report released by KPMG found almost 8 out of 10 Canadian students surveyed are using AI-generated content as their own work. But do the students consider this to be cheating? Two-thirds of these students said yes.

Between August to September 2025, KPMG surveyed 684 students attending university, college, vocational or high school educational institutions on their use of gen AI. KPMG is a full-service Audit, Tax and Advisory firm owned and operated by Canadians. (Graphic by SWCC)

Clearly, many students use gen AI, even if they feel that they are cheating, but why?

In a June 2025 Forbes magazine article, AI for Educators podcaster, Dan Fitzpatrick, interviewed San Jose high school student, William Liang, for his behind-the-scenes perspective on how his peers are using gen AI. Liang compares assignments to logistical hurdles that must be cleared as efficiently as possible. Gen AI gives him and his classmates a leg up on clearing hurdles through cognitive offloading, much like the calculators of decades past.

Is Gen AI Doing Too Much of Our Thinking?

Concerns about gen AI go far beyond simply cheating the system. Many wonder whether students will become overly reliant on gen AI at the expense of developing critical thinking skills. This question is also of interest to researchers.

For instance, a 2024 research study investigated the relationship between gen AI use and cognitive load. The study found that when university students used AI for research and answer-finding, their cognitive load was reduced. However, they also found that gen AI-use resulted in lower-quality reasoning for analysis and synthesis.

A 2025 study by Swiss researcher Michael Gerlich also looked at cognitive load. This time, it was in relation to cognitive offloading. Gerlich found that higher AI use led to greater cognitive offloading. He also found that, as cognitive offloading increased, critical thinking decreased. Especially notable was his finding that younger participants exhibited higher dependence on AI tools and scored lower on critical thinking compared to older participants.

(Graphic by SWCC)

So, given findings like these, how can we ensure that people use AI tools without losing their ability to think critically?

Brain First, Gen AI Second

A June 2025 study from the Massachusetts Institute of Technology looked at the neural and behavioural impacts of using a large language model (LLM) to assist with writing essays. LLMs are a type of gen AI that focuses mainly on generating text. Participants were divided into three groups. Each group used either an LLM, a search engine or their brain (no tools) to assist them during three essay-writing sessions.

Participants were connected to an electroencephalography (EEG) device to measure brain activity during the sessions. The EEG results showed significant differences in cognitive activity among the three groups. As tool use increased, cognitive activity decreased.

During a fourth session, the LLM group could no longer use tools, and the brains-only group could use an LLM. Brain activity was once again measured. The researchers found that the LLM-to-brain group struggled to quote their own work and felt little ownership over what they had written.

On the other hand, the brain-to-LLM group had high memory recall and the highest feelings of ownership of their work. It is insight from research such as this that can help educators effectively use gen AI with students.

Kim Taylor speaking at the 2025 SWCC Annual Conference on June 13th, 2025 in Fredericton, N.B. about the impact of generative AI on young students. Taylor was part of the "AI For Good" panel, alongside Concordia University PhD candidate, Azfar Adib. (Photo by SWCC)

The Bottom Line

Like the calculator, gen AI can be useful in reducing mental effort if foundational skills are in place. It can also be an effective tool when it is used to extend, rather than replace, cognitive effort. But research has shown that there can be some serious side effects. If gen AI becomes the default for human cognitive effort, critical thinking, as well as many other skills, will be at risk.

Above all, like any other tool, people need to know how it works. They need to know its strengths and limitations, when it can help, and when it can harm. Just because you have a hammer, it doesn’t mean that every problem becomes a nail.

Note: Only the built-in grammar and spell check of MS Word and human reviewers were used in the writing and editing of this blog.

A copy of this article can also be accessed at https://letstalkscience.ca/news-media/generative-ai-killing-critical-thinking

Further Reading/Listening

Brains Before Bots (New Research), AI for Educators Daily, Spotify.

On Gen AIin the Classroom: Give Up, Give In, or Stand Up, Active History.

The Metacognition Revolution: AI is Playing a Central Role in Reshaping How We Learn, The Atlantic.

About the Author 

Kim Taylor is the Strategic Projects and Integration Officer for Let's Talk Science; a national charitable organization committed to preparing youth for evolving career and citizenship demands in a rapidly changing world. She has been a passionate science communicator and informal educator for more than 25 years and enjoys finding ways to put the “A” in STEAM through sustainable fibre arts.