The risk of 'degeneration of thinking' due to AI misuse.

DNVN - Recent studies have sounded the alarm about the unpredictable consequences of humanity's increasing reliance on Artificial Intelligence (AI), especially large language models (LLMs) like ChatGPT.

Illustrative image.

AI is present in almost every aspect of human life, from daily activities to production, education , research, and artistic creation. AI helps people optimize operations, increase efficiency, and reduce manpower. The explosion of AI has demonstrated the continuous growth of machine learning and big data technology applications, pushing artificial intelligence to a new level of development.

Despite these enormous benefits, AI risks turning the very process of knowledge creation into a closed loop, threatening innovation and causing a 'degeneration' of human thinking.

According to experts, the immediate convenience of using AI may come at the cost of long-term intellectual resources.

Brain function decline and “cognitive debt”

A four-month study by the Massachusetts Institute of Technology (MIT) has shed light on the profound effects of AI overuse on the human brain. The study, which tracked 54 essay participants, found that reliance on AI tools can impair brain function and critical thinking skills.

Using electroencephalography (EEG) technology, researchers recorded the brain activity of the participants. The results were astonishing: The group using the AI ​​tool exhibited significantly different neural patterns compared to the group that only wrote mentally. This group showed up to 55% fewer neural connections than the brain-only group, particularly in areas responsible for deep thinking and memory formation.

Researchers call this phenomenon "cognitive debt." Similar to financial debt, which allows us to enjoy present benefits but incur costs later, cognitive debt occurs when we delegate intellectual effort to external systems, causing our mental muscles to atrophy from disuse. If abused, AI could turn humans into passive consumers, losing the ability to think independently.

The consequences of reduced cognitive function are clear:

Memory impairment : In the first session of the MIT study, as many as 83% of AI users were unable to cite from their own writing, and none provided an accurate citation. Even after several sessions, many still struggled with this basic task.

Weakening critical thinking : The use of AI leads to "superficial engagement" and weakens critical thinking skills, potentially fostering procrastination and laziness.

Reduced brain activity : Individuals who wrote essays with AI assistance activated significantly less brain activity, with areas related to memory, critical thinking, and executive function showing a sharp decline.

The crisis of knowledge creation and the "model collapse"

The problem with LLMs doesn't lie in their synthesis capabilities, but in the fact that they only know how to recycle what already exists. These systems don't generate new knowledge; they primarily use existing data and redistribute it.

Unlike humans, who are always seeking new directions for various motivations such as money, fame, curiosity, or a desire to conquer knowledge, AI cannot generate "novelty." When AI can answer almost any complex question in just a few seconds, the rewards for human creativity will gradually disappear.

This loss of momentum has been demonstrated by the case of Stack Overflow, a forum where programmers share knowledge: in just 6 months after ChatGPT appeared, the number of questions on the platform decreased by more than 25%, and has now decreased by 90%. This is the loss of a "living knowledge base," where every answer was verified, debated, and supplemented.

Data from platforms like Stack Overflow used to be the training source for AI tools. As this source of knowledge dries up, AI will increasingly repeat what it has already produced.

This phenomenon is known as "model collapse." Researchers warn that when models are trained primarily on data generated by the AI ​​itself, the quality of knowledge gradually deteriorates, losing its ability to reflect the diversity and accuracy of the real world . Professor Hannah Li of Columbia University concludes: "The overall result is that they will get worse. The models will get worse."

The risks of knowledge homogenization and recycling.

Besides the decline in individual awareness, the misuse of AI also leads to the homogenization of knowledge. Essays written with the aid of AI show striking similarities, revealing "statistically homogeneous" patterns. This creates a kind of intellectual monoculture, where diverse human perspectives are filtered through the same algorithmic lens, potentially stifling creative diversity.

AI tends to emphasize common patterns in data, gradually overlooking rare information, unique details, or data hidden in the shadows of statistics. If this trend continues, the Internet, a rich repository of information, risks becoming a diluted world where knowledge is lost. Humanity may enter an era of "knowledge recycling," where innovation stagnates as the rewards for new creations become increasingly obscured.

AI is a powerful tool, but it requires a strategic approach. MIT researchers have found an encouraging sign: participants who had developed a solid cognitive foundation before using AI were able to leverage it more effectively, with even increased neural activity upon introduction. They used AI as a true assistant, not just a support tool.

Human knowledge is built through effort, curiosity, and the courage to venture into the unknown. However, if we turn AI into the sole "railway," we risk being led into a vicious cycle where both humans and machines only repeat what has already been done.

Hien Thao

Source: https://doanhnghiepvn.vn/cong-nghe/nguy-co-thoai-hoa-tu-duy-vi-lam-dung-ai/20250928033804291