Question

ChatGPT is ruining your memory.

Answer 1

This claim is largely true because numerous psychological studies and reports have concluded that excessive dependence or usage on ChatGPT leads to poorer memory, lower brain engagement, and a loss of cognitive ownership. A study from MIT, found in Psychology Today, showed that in simple tasks, such as writing an essay, the AI assistant of ChatGPT lowered students' creativity due to the lack of effort, which lowered other things, such as critical thinking and overall growth. An article by Jordan Gibbs in the Medium shares a similar idea that AI helpers such as ChatGPT "poison" one's brain. 

It's important to know there is direct evidence that correlates with the use of AI engines like ChatGPT to have negative effects on education and overall brain activity, including memory. It's also important to note, however, that while research has shown negative effects of overusing ChatGPT for cognitively challenging tasks such as school work, AI, still being a relatively newer concept, isn't backed up by the most extensive research to prove a cumulative change over time. In MIT's study, which is largely referenced in this debate, the researchers themselves said, "We had a limited number of participants recruited from a specific geographical area, several large academic institutions, located very close to each other..." And for the grading of the essays written by students, they had a specially built AI agent, showing that there are still benefits and resourceful ways to use AI that aren't necessarily harmful. This is still an ongoing debate, but there is evidence that an overuse of AI assistance can affect memory negatively.

https://www.brainonllm.com 
 

Answer 2

There is limited evidence out there which supports this claim, but the evidence that is available points towards this claim being true. 

In a study from MIT headed by Nataliya Kosmyna, participants were divided into three groups: LLM-assisted, search-engine assisted, and brain-only. These groups each completed three essay-writing sessions under the same conditions. MIT used an EEG to assess cognitive load in each participant. Their results found that brain-only users exhibited the strongest and most distributed networks, with brain activity scaling down with use of tools, LLM's giving the most assistance and yielding the least brain activity. Over the course of four months, LLM users consistently underperformed on neural levels. During this study, LLM users and brain-only participants were asked to rewrite what they had just wrote without LLM assistance. Participants who used LLMs in the first writing had a difficult time recalling what they had just wrote, while brain-only participants were able to rewrite with little difficulty (Time). This study has not yet been peer-reviewed, but early evidence points to clear decline in cognitive activity with the use of LLMs.