Researchers from MIT and other institutions are investigating how generative AI influences human memory and cognitive abilities. Experiments show that AI-generated images and text can implant false memories, making individuals confident in events that never occurred, such as non-existent political scandals or movie remakes.
A study involving 200 participants demonstrated that those exposed to AI-altered photos were more likely to develop false memories and expressed higher confidence in them. Another experiment with 331 participants revealed that even when warned about potential errors, people often accepted AI-generated misinformation regarding video content, significantly impairing their recall accuracy.
Cognitive scientists are concerned about 'skill atrophy,' where reliance on AI for tasks like writing or medical diagnosis reduces human performance. A 2025 study in The Lancet Gastroenterology and Hepatology found that doctors' ability to independently detect precancerous lesions dropped by approximately 20% after regular AI use.
Research using EEG monitoring during SAT-style writing tasks showed that participants using ChatGPT exhibited weaker neural connections compared to those writing independently. Furthermore, those relying on AI struggled to recall the content they had just produced, suggesting that generative tools may bypass the mental effort required to consolidate memory.
While some researchers, like MIT professor Pattie Maes, are developing AI 'memory assistants' to help patients with dementia maintain independence, the broader implications remain debated. Experts warn that human oversight may be ineffective if AI fundamentally alters the cognitive processes used to verify information, potentially rendering traditional safeguards obsolete.
