Your brain works differently when you are using Generative AI for a job when you use your brain alone. That is, you are less likely to remember what you did. This is the result of a clear voice of MIT study, which has seen how people think when they write the article.
StudyA print print that has not yet been reviewed, it is very small (54 participants) and initials, but it indicates how our brain functioning uses tools like Openi’s chat GPT. Openai did not immediately respond to a request to comment on research (disclosure: CNET’s parents company, Zef Davis, filed a lawsuit against the Open in April, alleging that he violated Zef Davis copyrights in training and running his AI system).
These results show a significant difference in what happens to your brain and your memory when you complete a task using an AI tool instead of when you just do it with your brain. Researchers say that but don’t read too much in these differences-this is just a glimpse of mental activity at this moment, not how your brain runs all the time, there is no long-term evidence of changes.
“We want to take some steps in this direction and encourage others to ask questions as well,” Natalia CosmeenaA research scientist and the main author of the study, told me.
The growth of AI tools such as Chat Boats is changing rapidly how we work, searching and writing information. All of this has happened so fast that it is easy to forget that at the end of 2022, a few years ago, a few years ago, had emerged as a famous tool. This means that we are just starting to see research on how the use of AI is affecting us.
Take a look at what MIT studies about what happened in the brains of Chat GPT users, and take a look at what we can tell us from future studies.
See this: Testing Openai’s new chat GPT search engine
It is your mind on Chat GPT
Researchers from MIT divided their 54 research participants into three groups and asked them to write articles during a separate session within several weeks. One group was accessed to Chat GPT, the other was allowed to use a standard search engine (Google), and in the third there was none of them, only they had their own brains. Researchers immediately wrote articles. After interviewing the articles later, researchers interviewed articles, and recorded the participants’ mental activity using electrolysis, or EEG.
The language analysis used in the articles showed that people in the “Mental Mental” group wrote more different ways, while large language models produced very similar articles. After writing articles, the interview yielded more interesting results. People who used their brains alone show better memories and better than their writing than those who used search engines or LLMs.
It is not surprising that the more heavy trustworthy on the LLM, who could have copied and pasted from a chat boot response, would be less able to refer to what he wrote. Cosmeena said that this interview was written immediately. Later, the lack of memory is noteworthy. “You wrote it, no?” He said. “Don’t you know what it was?”
The results of the EEG also showed significant differences between the three groups. The interaction between brain components was more nervous contacts-a minimum activity in the and LLM group among the brain participants compared to the search engine group. Once again, not a completely amazing result. The use of tools means that you use less than your brain to complete a task. But Cosmeena said that this research helped show what the differences were: “The idea was to look closely to understand that it was different, but how is it different?” He said.
Natalia Cosmine shares a photo of a research article when an EEG monitors brain activity.
The authors of the study wrote that the LLM group “reduced weak memory marks, self -monitoring and scattered writings.” This may cause concern in the learning environment: “If users relies heavily on AI tools, they can gain levely flow, but fail to make knowledge internal or feel owned by it.”
After the first three articles, researchers invited the participants back for the fourth session in which they were assigned to a different group. The results from the significantly small group (only 18) of the subjects showed that the use of LLMs also showed more activity by those in the brain group, while the LLM only showed less nervous contact without LLM than the initial mental group.
This is not ‘BRINTEROT’
When the MIT study was released, many headlines claim that it shows that the use of chat GPT is causing the brain “rot” or long -term problems. The researchers did not find the same, Kosmina said. The study focused on brain activity that participants were working – at that moment their brain’s internal circuitry. He also reviewed the memory of his work at that moment.
Long -term studies and different methods will be needed to understand the long -term effects of AI use. Cosmeena said future research can look at other general AI use cases, such as coding, or used technology that tests different parts of the brain, such as functional magnetic resonance imaging, or FMRI. “The whole idea is to encourage more experiences, collecting more scientific data,” he said.
Although the use of LLMS is still being researched, it is also likely that the impact on our brains is not as important as you can think, said Jennio Stein O’Brien, assistant professor of neuro science at John Hopkins University, who was not included in the MIT study. She studies how genetics and organisms help develop and build the brain – which happens in early life. He said that these delicate periods stop during childhood or youth.
Stan O’Brien told me, “All this happens before you ever talk to Chat GPT or anything like this.” “There are a lot of infrastructure that has been compiled, and it is very strong.”
Stan O’Brien said the situation may vary in children, who are rapidly in touch with AI technology, though children’s studies raise moral concerns for scientists who want to research human behavior.
You can help you write an article, but do you remember what you write?
Why do you care to write an article anyway?
The idea of ​​studying the effects of AI on writing the article seems meaningless to some people. However, was it not a matter of writing an article in school to get a grade? Why not outsource this job for a machine that can do it, if not better, then more easily?
The study of MIT reaches this task: writing the article is about promoting your thinking, about understanding the world around you.
“When we start writing, we know that we start the next questions and think about new ideas or new content,” said Robert Kumings, a professor of written and rhetoric at Mississippi University.
The same research has been done on the way we write computer technologies how we write. A study The completion of the sentence included in it – which you can know informally automatically. He took 119 authors and entrusted them with the task of writing an article. About HALF half of computers were present with Google Smart Composes, while the rest were not. Did he make the authors faster, or did they spend more time and write less because they had to go to the proposed elections? The result was that he wrote about the same amount at the same time. He told me, “He was not writing in different sentences, different levels of complexity of ideas.” “It was straightforward.”
Read more: According to our experts, AI accessories: 29 ways to work for General AI for you
Chat GPT and its ILK are a different animal. With the technology of completion of the sentence, you still have control over the words, you still have to make a written choice. In the study of MIT, some participants just copied and pasted about chat jeptic. They may not have read their work on their own work.
“My personal opinion is that when students are using Generative AI to change their writing, they are a surrender, they are no longer actively engaged in their project,” said Kums.
MIT researchers found something interesting in this fourth session when they saw that the group, which wrote three articles without tools, had a high level of engagement when the tools were finally given.
He wrote, “together, these results support an educational model that delays AI integration unless learners are engaged in sufficient academic efforts.” “Such a point of view can immediately promote both the utility of the device and the lasting academic sovereignty.”
Kumings said he had started teaching his composition class without any devices. Students write in the class hand, usually on topics that are more personal and it is difficult to feed in LLM. He said that he does not feel that he is classifying the papers written by AI, that his students are getting the opportunity to engage with their thoughts before seeking help from a device. “I’m not going back,” he said.