Is ChatGPT Making You Smarter? 83% People Can’t Recall What They Wrote Using It

Boston, MA : A groundbreaking study conducted by MIT’s Media Lab has raised serious concerns over the cognitive and environmental impacts of relying on Large Language Models (LLMs) like ChatGPT. The study, involving 54 participants aged 18 to 39, compared essay-writing outcomes using ChatGPT, Google Search, or no digital aid. The results were startling.

Quoting Ability Breakdown

The research revealed that 83.3% of the ChatGPT-assisted group were unable to quote from essays they had written just minutes earlier — a sign of weak neural engagement and poor memory consolidation. In stark contrast, only 11.1% of participants using search engines or relying solely on their brains faced similar difficulties. EEG scans confirmed these findings, showing the lowest brain engagement levels in the ChatGPT group.

Ownership and Satisfaction

While the AI-assisted group produced grammatically polished essays, participants reported lower satisfaction and a sense of detachment from their work. Essays were described as “predictable” and “soulless” by both human and AI evaluators. On the other hand, brain-only participants demonstrated deeper vocabulary, originality, and critical thinking. Notably, the search-assisted group reported the highest levels of satisfaction and ownership.

The Cognitive Cost

Researchers warn of a phenomenon they’ve dubbed “cognitive debt” — a long-term decline in independent thinking caused by over-reliance on AI tools. “It’s like trading long-term brain capacity for short-term efficiency,” the study notes. In later sessions, ChatGPT-dependent users struggled when asked to write without assistance, whereas the brain-only group showed improved performance once introduced to the tool — suggesting AI is best used after active thinking.

Productivity vs. Reasoning

Although ChatGPT made participants 60% more productive by reducing extraneous cognitive load, it simultaneously reduced germane load — the kind essential for deep learning and reasoning. Those using traditional search methods produced higher-quality arguments in scientific writing tasks.

Environmental Toll

The study also highlighted an alarming environmental disparity. A single LLM query consumes 0.3 Wh — ten times more than a standard Google search. Over 20 hours, LLM use tallied 180 Wh, compared to just 18 Wh for search engines. With LLM usage rising, the researchers warned this ecological cost may soon fall on users.

Final Verdict

While ChatGPT improves grammar and speed, it appears to come at the cost of originality, deep thought, and even neural engagement. The message from MIT researchers is clear: AI should augment human thinking — not replace it.

GPT and similar AI tools should be seen as aids to enhance and speed up our work, not as substitutes for human thought, creativity, or presence. It is important that companies and individuals use AI responsibly—complementing human effort, not replacing it.

As Holy Father Leo XIV, shares this thought in the recent jubilee year events and calls AI an exceptional human creation but insists it must remain a tool, not replace human dignity. He stresses that AI should help, not hinder, the development of children and youth. The Pope warns that access to vast information is not the same as true intelligence or wisdom. He urges that AI be guided by ethics, promoting the full material, intellectual, and spiritual growth of the person. True wisdom, he concludes, lies in understanding life’s deeper meaning, not just processing data

Source : Insights by Inshort