The truth behind ChatGPT brain-rot allegation
MIT researchers found that utilizing ChatGPT to compose essays can cause ‘cognitive debt’ and a ‘probable decline in learning skills’. There is significant concern that the use of AI may result in a widespread “dumbing down” or a decline in critical thinking skills. If students use AI tools prematurely, the argument suggests, they might fail to cultivate essential skills in critical thinking and problem-solving. Since the emergence of ChatGPT nearly three years ago, the influence of artificial intelligence (AI) technologies on education has sparked extensive discussion.
Are they useful tools for personalized education, or pathways to academic dishonesty? The study’s findings were significant. The distinction between employing AI and relying solely on the brain, over four months, the MIT team asked 54 adults to write three essays using AI (ChatGPT), a search engine, or their own brains (“brain-only” group). The team assessed cognitive engagement by analyzing electrical activity in the brain and conducting a linguistic analysis of the essays.
The cognitive engagement of AI users was significantly lower than that of the other two groups. This group struggled more with recalling quotes from their essays and experienced a diminished sense of ownership over them. Participants switched roles for a final, fourth essay, with the brain-only group utilizing AI and vice versa. The AI-to-brain group exhibited poorer performance, with engagement levels only marginally surpassing those of the other group during their initial session, significantly trailing behind the brain-only group in their third session. The authors assert that this illustrates how extended use of AI resulted in participants accruing “cognitive debt.” When given the chance to utilize their intellect, they could not match the engagement or performance levels of the other two groups. They cautiously observe that only 18 participants (six per condition) completed the fourth and final session. The findings are preliminary and necessitate additional testing.
These results do not imply that students who used AI incurred “cognitive debt.” The findings stem from the specific design of the study. The alteration in neural connectivity within the brain-only group during the initial three sessions was probably due to increased familiarity with the study task, a phenomenon referred to as the familiarisation effect. As study participants repeat the task, they grow more familiar and efficient, and their cognitive strategy adjusts accordingly. When the AI group finally had the opportunity to “use their brains,” they were only performing the task once. Consequently, they could not match the experience of the other group. They recorded marginally improved engagement compared to the brain-only group in the initial session.
To fully substantiate the researchers’ claims, the AI-to-brain participants must also engage in three writing sessions without AI. The brain-to-AI group’s more productive and strategic use of ChatGPT can likely be attributed to the fourth writing task, which involved composing an essay on one of the earlier three topics. Writing without AI demanded greater engagement, leading to improved recall of past writings. They primarily utilized AI to search for new information and refine their previous writings.
To grasp the current landscape of AI, we can reflect on the introduction of calculators. In the 1970s, their influence was controlled by increasing the difficulty of exams. Students were expected to use calculators instead of doing calculations by hand, allowing them to focus their cognitive efforts on more complex tasks. The bar was significantly raised, leading students to work as hard, if not harder, than they did before calculators were available. The challenge with AI is that, largely, educators have not elevated standards to the extent that makes AI an essential component of the process. Educators continue to require students to complete the same tasks and expect the same standard of work as they did five years ago. In such situations, AI can be harmful. Students largely delegate critical engagement with learning to AI, leading to “metacognitive laziness.”
AI, much like calculators, can and should assist us in achieving tasks that were once deemed impossible, while still necessitating substantial engagement. For instance, we could instruct students to utilize AI to create a comprehensive lesson plan, which will subsequently be assessed for quality and pedagogical effectiveness during an oral examination. In the MIT study, participants using AI produced the “same old” essays. They modified their engagement to meet the expected standard of work. Students would experience the same outcome whether they were asked to perform complex calculations with a calculator or without one. The group calculating by hand would sweat, while those using calculators would hardly react.
Current and future generations must develop the ability to think critically, creatively, and solve problems. AI is redefining the meaning of these concepts. Writing essays by hand no longer showcases critical thinking skills, just as performing long division no longer reflects numeracy. Understanding when, where, and how to utilize AI is essential for sustained success and skill enhancement. Determining which tasks can be delegated to AI to alleviate cognitive debt is equally crucial as recognizing those that demand true creativity and critical thinking.
