Fifty-four participants from top universities like MIT, Harvard, and Wellesley were divided into three groups:
- The LLM Group: Could only use ChatGPT to write their essays.
- The Search Engine Group: Could use Google, but not AI.
- The Brain-only Group: No tools. Just their own thoughts.
After each 20-minute writing session, the researchers asked a simple question: “Can you quote any sentence from the essay you just wrote?”
The answers were unbelievable
In the first session, a staggering 83.3% of ChatGPT users failed to quote a single sentence correctly.
Meanwhile, the Brain-only and Search Engine groups reported their work with almost perfect accuracy. They had completed the mental effort, and the memory served as the reward.
This was not just a one-time fluke. The LLM group’s poor performance in memory and recall was a consistent issue throughout the entire study. They were disconnected from the very work they were producing.
