_ AI in Writing
Using AI as a go-to tool for research and writing can impair critical thinking, impacting our capacity to learn, ability to communicate, to make decisions, or to solve problems.
When researching for a topic or presenting these findings in a cohesive summative essay, using AI & LLMs can negatively impact our cognitive engagement and reflective thinking. In a recent study comprising a survey of a large group of knowledge workers, it was found that individuals were far less engaged in critical thinking to complete their tasks while placing greater confidence in generative AI, and became onlookers of AI’s production rather than active participants in solving the problem at hand. Relying more on the AI’s accuracy rather than their own reasoning led to a decline in their cognitive effort and reflexive scrutiny.
Another study based on systematic EEG scans compared three groups of students: LLM users, search tool users, and brain-only users for the same writing tasks. Participants who used an AI assistant repeatedly showed lower neural engagement, especially in areas linked to attention, planning, and memory, compared to those working without it. The study found that their reduced brain activity was further evident in weaker ownership of ideas, failure to recall information from their own work, and lower-quality work when AI assistance is not available.
As we delegate the tasks of research, summarization, and reasoning to AI, we reduce our own mental effort and engagement in the whole process of synthesizing information, weighing evidence, or structuring arguments. This is called cognitive offloading, where the brain’s muscle for critical thinking is saved from its necessary exercise. Our role shifts from a thinker with insights to an effective prompt designer and, evidently, an editor of AI outputs, only superficially engaged with content instead of comprehending it.
And through neural evidence, we understand this reduction in mental effort is not just behavioral. Lower engagement during critical thinking tasks among habitual AI users is measurable in brain activity. Lower brain activity weakens the neurological pathways related to attention, memory formation, and executive function. These processes are related to problem solving, decision making, and, evidently, to independent thinking. That, in turn, inhibits the ability to learn and communicate effectively.
We are already using many tools as our cognitive partner in a learning environment, from Google to a calculator. Still, generative AI and LLM have diminished most effort on our part, and it is also diminishing the need to diagnose or discern, differ or disagree, deduce or reason, and then to reflect, create, or invent. Lesser need will result in lesser activity, lesser stamina, and lesser ability.
References:
