AISeptember 15, 202610 min read

Is AI Eliminating Critical Thinking?

What’s the difference between a human like myself writing this newsletter and a chatbot writing it?

What’s the difference between a human like myself writing this newsletter and a chatbot writing it? According to popular culture and economic improvement, there is no difference, and in fact, it’s celebrated as an efficient use of time and resources! The AI chatbot can write the exact same material, in the same voice as the user, with the same statistics, and free up the user to do “more important tasks” while it’s spitting out paragraphs in seconds. So what’s the big deal? So what if I want AI to just finish a homework assignment really quick, I’m not even interested in the class! I don’t really want to do that report, so I’ll just put down some thoughts and let the AI do the rest with context, that should be enough right?

Wrong! If we as humans start delegating more tasks away from our cognitive functions to AI, we will weaken our critical thinking and decision making skills in our personal and professional lives. Right now, we are seeing a reliance on technology as never seen before with very little awareness regarding the effects it has on the human mind. In today’s edition, we are going to be talking about what is currently being seen regarding AI and the human mind and how we can improve the degradation of our critical thinking abilities.

The Effect of AI on Critical Thinking

There was a study done by Microsoft researchers and Carnegie Mellon that sought to answer the question “When and why do knowledge workers perceive increased/ decreased effort for critical thinking due to GenAI?” (Hao-Ping Lee, ACM Digital Library 2025). Now for those who don’t know or haven’t read previous editions of this newsletter, GenAI (or Generative AI) is a type of artificial intelligence that can create original text, images, code, or audio/video based on the prompt that a user puts in. Chances are that most people reading this are actively using Generative AI every day, which there’s nothing wrong with, but it goes to show the level of use that GenAI has right now and why this study is so important. Most people are using it, therefore most people could be experiencing the same results as the given study.

The study collected 936 real world examples and 319 “knowledge workers” of people using Generative AI at work. It was found that workers with MORE confidence in AI were associated with less critical thinking ability than someone who had less confidence in AI. People who were more confident in their own ability, generally had more critical thinking ability than those relying more heavily on AI. Not only is this new reliance changing how people think, but it’s changing where people think as well. In the same study, most workers using AI had to use critical thinking skills, not on the actual output, but on the sourcing of information gathered from the Large Language Model. Those knowledge workers were checking information from AI, rather than sourcing it themselves and thinking of how it applies in their work in the process.

Now, you may look at this and say “Well duh, Matthew. Of course if someone is leaning more on AI, they won’t think as critically.” While this seems obvious, there is something extremely fascinating about this discovery. If someone told you using common phrases such as “If you use this tool a lot, it’s going to be worse for your brain and your own ability to learn, but you’ll finish quicker”, are you immediately jumping on the bandwagon? Now, imagine that same person told you “If you prompt very specifically and check the sources that this tool used, it’s the same thing as just getting the information yourself”, I imagine you would be more inclined to use it. I would, absolutely! Who doesn’t want to save time and if there are little consequences, why not?

The problem is, the process doesn’t stop with just prompt engineering and then getting the information needed. The study found that increasing reliance on AI was actively damaging a person’s ability to critically think and found that cognitive fatigue was common. This means that the workers more inclined to trust AI were unlikely to think about a problem for a longer period of time than someone who trusted in themself more. This is a bad omen for the upcoming generations and workforce, where decision making and critical thinking is now a cornerstone for companies hiring new talent. (World Economic Forum, 2025)

People Are Now Prone to Accept AI as Fact

In another study, researchers had three test groups. One, where there was no AI help. Two, where there was faulty AI help. Three, where there was faulty AI help, but there was a warning about the faultiness. All three groups were tested on the same Cognitive Reflection Test (CRT) and were told to answer all of the questions as correctly as they thought with whatever help they had. The study found that the group who had faulty AI help answered “fewer than half as many questions correctly as the group receiving no AI help.” (Science Direct, 2025)

Again, you might think this is obvious, but look to the other group who had AI help but had a warning to double check their work. They scored nearly DOUBLE the correct amount of questions that the group who didn’t have the warning did with faulty AI. However, Group Three still did not match the amount of correct answers as the ones with no help. This shows that just because people are more experienced with AI or are still double checking their work, doesn’t mean that it’s better than just completing the work yourself with your own critical thinking skills.

Humans that use AI in their daily lives, are now more prone than ever to accept whatever the AI is giving them as “fact” for a variety of reasons. One, we have been led to believe by the makers of these LLMs (OpenAI, Anthropic) that these models are state of the art and are designed to increase productivity and effectiveness. We (those who use their models) generally associate that with the statement of “It works, use it”. When I buy a product, I tend to believe the people who are selling it to me of its reliability and effectiveness, I don’t tend to question it. With AI now, we are told it’s better than ever before and all of the capabilities around it, leading us to be more accepting of relying on its use.

Other reasons are popular opinion and the “herd mentality” of just following what others are saying. When everyone is telling a person how great AI is and they should be using it more, without the proper ramifications of how to use it effectively, those people are more likely to trust what AI is saying without double checking the work. If your job requires you to use AI more and are expected to radically change how you do your work, someone would be more likely to change their previous routines and trust in LLMs for their performance.

All of these things are happening right now in the modern business world.

“Automation Bias”

The Organization for Economic Co-Operation and Development (OECD) did their own analysis on AI in the workplace and found that government institutions using AI in their work are actively accepting their results as neutral and impartial. (OECD, 2025) The official word for this is “automation bias” and it is the danger of people accepting what an AI chatbot has to say simply because it is the system and giving it more objectivity than our own thoughts. There are different types of automation bias that I will give brief explanations to here.

There is ‘Commision Errors’ which is when a system gives a set of instructions of how to complete a task and the user does it even though there is evidence suggesting that there is another way to do it. Then there are ‘Omission Errors’ when the system does not flag something so the user neglects checking if there was an error done by the system because they trust it. Researchers found that humans are much more likely to want to have systems to do automated work and help them achieve “the road of less cognitive effort”. (Sage Journals, 2024)

So, with this information, people are more interested in automating a task and completing it for them, whether it’s gathering information from websites, writing out an article, or coming up with ideas of what to do next. This is all actively destroying the critical thinking skills of the working mind and there needs to be some sort of awareness around this.

The Effects on Students

So far, we’ve only been talking with the assumption that we are referencing working adults that have presumably graduated from college and high school. Now we are going to talk about the effect it has on students right now. Even though there is not much data and research on AI during the development of student’s brains, there was one done very recently that highlights the real impact AI has on the brain.

In 2025, Acta Psychologica did a study on the observance of 580 college students to see if a higher dependence on AI led to lower critical thinking levels. The study concluded that the correlation between higher dependency on AI and critical thinking skills is r=-.35. This negative correlation shows the effect that Artificial Intelligence is having on college students in America. (National Library of Medicine, 2025)

Now, this doesn’t mean exactly that AI caused this because this was simply an observational study, but the amount of data we’ve seen so far seems more than likely that this was the effect of too much AI dependence for tasks. It’s a question now of the balance that schools must take in order to properly prepare the next generation of students for the workforce. Students need to be taught AI responsibly and be given the facts of the side effects of using AI too much as well as teaching them the possibilities and opportunities for using AI effectively. I would like to see, personally, more courses teaching how to use it more effectively and responsibly rather than doing nothing or banning it completely in the classroom which seems like a trend happening in the country right now. (Tech Policy Press, 2026)

Why I’m Optimistic

I know I’ve said a lot of really unfortunate and almost doomsday predictions about the negative effects of AI on future generations. But, I really do think that Artificial Intelligence will make everyone better off overall and increase the wealth and standard of living for millions of people across the globe. Medical researches using AI have found more possibilities for cures of diseases (BBC, 2026) than ever thought possible. Corporations are using AI to find cheaper ways of producing and distributing goods/services across the globe, making it cheaper for the average consumer.