Psychology of Singularity – Your Source for AI Information

Image by Hotpot.ai

Is AI Making Students Lazy and Dumb?

By Jill Maschio, PhD

Educators across America are concerned about their students’ lack of motivation to work hard when using AI. Students’ test scores appear to be declining, and there is growing evidence that intelligence is decreasing. Is AI in the classroom making students less motivated to read, write, and learn, or making them more creative and curious about learning?

As AI becomes more popular, we must realize that not all trends are equal. Some trends are helpful, while some are harmful or pretty amusing. We don’t understand the whole story about AI and how it works, according to Sam Altman (Curry, 2024), who addressed questions about AI at an A.I. conference. Whether we know the whole story about AI, education is scrambling to find practical uses in the classroom without first knowing whether there is any real benefit to learning. If you have read my article about AI and its impact on Intelligence, you can access it here.

We need a better understanding of the potentially harmful and beneficial effects of AI in education. We would assume there are benefits, much as society and educators believed there would be huge gains in learning when public schools first introduced computers to kindergarten classrooms. Later, we learned of some harmful consequences of excessive Internet use. An example of the harm AI might have on academic performance has yet to be thoroughly researched. Marcello Mariana (Dwivedi et al., 2023) writes that generative AI has not only been inaccurate at times but also leaves humans without independent thought. Its use can lead to meaningless learning for humans and innovation. If AI designers could consider that learning needs structure, it may be even more useful in the classroom. You can read more about that here. 

In a two-part study by Abbas et al. (2024), researchers examined the effects of ChatGPT on procrastination, memory loss, and academic performance among 494 students. The participants reported on a self-reporting instrument about their procrastination level and memory loss, such as being unable to retain information as much as they used to. The researchers collected their latest reported CGA to examine memory loss. They reported that memory loss is associated with excessive use of ChatGPT for academic tasks. The continuous use of AI may lead students to develop a lazy attitude towards learning.

In another study by Petric (2024), students’ grades improved when they used ChatGPT. A closer look at the study reveals that students used the chatbot to prepare for exams and pre-exams, produce entire homework assignments, write seminar papers, write part of a project, obtain advice on how to improve a test or assignment, verify the accuracy of their work and calculations, obtain critical evaluations of completed assignments, acquire assistance in improving language style and proofreading, etc. Academic performance in the study is assessed through self-reported grades or perceptions, which may not accurately reflect actual academic performance. Incorporating objective measures, such as GPA or standardized test scores, would strengthen the validity of the conclusions. As-is, the study does not appear to compare the use of ChatGPT to other study methods or whether ChatGPT was used in conjunction with other methods. Due to this issue with the study, one cannot assume that ChatGPT improves academic performance. The field needs more peer-reviewed research to help us identify if AI truly fosters academic learning and intellectual growth by comparing actual learning against some other factor(s).  

Theory of AI and Reduction in Metacognition

I hypothesize that students who lack strong metacognition may not be aware of what they know and what they don’t, particularly when prompted by generative AI (e.g., ChatGPT). While people eagerly want information at their fingertips, they can easily ignore the process of reflecting on what was learned and what was not, and identifying the focus to understand the content. If this hypothesis is true, then overreliance on generative AI tools for learning may reduce memory formation because active cognitive engagement with information is necessary for selective priority information to be consolidated into long-term memory (Cowan et al., 2021).

The Use of AI may Limit Neurological Activity Needed for Memory Formation

Nature has predetermined that if a specific region of the brain cannot store information, then the brain cannot be modified by experience, leaving the individual unable to remember (LeDoux, 2002). Memory formation is complex. It involves neurochemicals, for instance. When first exposed to stimuli, two neurons begin to fire together, and the neurotransmitter glutamate helps strengthen the neural connections. If the firing neurons are reactivated, their dendrites begin to physically grow, establishing a neural network.

However, factors may influence the neural network’s stability. According to Shors (2014), deep processing of information can help stabilize neural connections. The duration of a neural networkis activity may also help stabilize the neurons. For example, the longer we spend engaging with content, the more the neural network stays active – potentially increasing its stability.

So, what is knowledge acquisition?

It is the process of acquiring new knowledge.  

How is knowledge gained?

Knowledge is gained through the processing of information, which involves encoding it (through sensory systems, whether we read or hear). Information must be encoded, organized, and form meaning before consolidation into long-term memory. Working memory is significant for learning, helping people comprehend new information.  

An excerpt from Kasneci and colleagues’ (2023) article states how AI in the form of LLMs can be used in higher education. 

“ For university students, large language models can assist in the research and writing tasks, as well as in the development of critical thinking and problem-solving skills. These models can be used to generate summaries and outlines of texts, which can help students to quickly understand the main points of a text and to organize their thoughts for writing. Additionally, large language models can also assist in the development of research skills by providing students with information and resources on a particular topic and hinting at unexplored aspects and current research topics, which can help them to better understand and analyze the material” (para. 10). 

They are correct, but consider that by having an LLM write summaries for them, the learner is not engaging in the process of figuring out what the author is saying. It is, at times, through the process of trying to determine what is being said and using the working memory to comprehend and form relationships between ideas that greater learning is the end result. And wading through the research to identify gaps and research ideas is part of the learning process. It is what helps turn a novice into an expert. The learner processes information deeply and makes meaning of it, which is one way the brain learns naturally. Having to process information fosters independent thought. 

McGehee (2024) examined the correlation between grades and the use of AI tools among 2,154 students in grades 9-12. Students’ use of AI was grouped into two categories: tool and facilitator. Using AI as a tool included conducting research and finding information, summarizing information, and acting as a writing and editing assistant. AI as a facilitator included explaining concepts and principles in easy-to-understand language, creating study guides and sample test questions, and a tutor or teacher. The research shows a weak correlation between using AI as a tool and a facilitator, but not when using AI for either. The results were significant for English, health, PE, and social studies, but not for mathematics, science, world languages, or other studies. 

The weakness of this study is that it fails to show whether learning actually occurred (no cause-and-effect) or whether grades were correlated with the use of AI when students used it as a means of personal assistance and for writing and editing. Furthermore, the use of AI was not compared to other forms of assistance or methods of studying. Because there was no comparison to other forms of studying or assistance, there is nothing to compare the results to. 

Some Best Practices for Using AI in the Classroom

• The LLM does not triumph over humans in the learning process simply because it can summarize faster.  
• The benefits of using AI in the classroom should outweigh the risks.
• The learner should engage with AI so that deep processing occurs.
• AI should not be used to simply reduce the workload or to ease it, unless the workload is outside of the students’ skill level. For example, if you are teaching a social science course, then having students create a app using an AI app maker where coding is not required, students can have the opportunity to create an app by apply the concepts you are teaching.
• Do not apply technology simply because it is trendy.
• The technology (AI) should give the learner opportunities to think deeply by analyzing, evaluating, and synthesizing information, or creating new information or ideas. 
• The technology (AI) should help foster learning.  

A Theoretical Base for Using AI in the Classroom

There are many learning theories. Jean Piaget believed that learning is an accumulation of knowledge where the learner creates their own reality. Without embedding AI into a learning theory, it’s pretty much hit-and-miss. The educator is simply using another teaching tool without a particular purpose. I propose that educators use AI (LLMs) to help students move beyond being a novice and up the continuum towards becoming an expert. You can read my paper Reassessing Educational Goals for Learning, Assessments, and Outcomes-How Education can Embrace AI for Student Success and Humanity’s Intelligence for more information about that. Humans will need to contain AI in the near future, and part of that entails that AI needs human input.  

When AI Use Influences Loss of Memory Formation

Depending on how a person engages with AI, it may be comparable to using a search engine to locate information that we glance at or to scrolling on our cell phones, both of which may reduce the time spent processing information, because if two neurons began firing, the connection would not become stable by moving on to other tasks or scrolling.

There is much more research to consider about this topic. All the research in the literature was not mentioned here, and I’m confident that generative AI tools are useful for learning. I do believe that how a person uses AI in the classroom makes the difference, much like my example above about using search engines to locate information. We can use a search engine to find information, then decide what to do with it. People can either simply forget it or use the information by processing it further into something meaningful.

Note: A human wrote this essay. No AI tools were used in the process of writing this article.

Note: A human wrote this essay.