Artificial intelligence is no longer an obscure term from science fiction.
In the last decade and especially since the launch of ChatGPT at the end of 2022 it has become a daily tool accompanying us in writing emails composing assignments editing images preparing presentations and even making personal decisions.
Corporations like OpenAI Google Anthropic and Meta are advancing in the race for advanced models each promising growing capabilities smarter answers and higher user friendliness.
And when these tools are available to everyone it is not surprising they also become an integral part of learning work and thinking itself
The advantages of using artificial intelligence are quite clear A study published by researchers from MIT and Stanford in 2023 showed that people who used GPT in writing tasks completed the task faster in forty percent less time and also received higher grades for clarity and quality.
Another study focused on software development teams found they worked up to fifty six percent faster when using the AI assistant in programming tools like git.
In other words it is a tool that increases productivity improves output and streamlines processes
But not everything is positive A study published on a preprint website by researchers from the MIT Media Lab examined for the first time what happens to our brain when using large language models like ChatGPT and the results are concerning.
The researchers examined fifty four participants performing writing tasks in three scenarios: without any aids while using a search engine and with GPT.
At the same time they were measured using EEG a technology that examines the brain’s electrical activity and also examined for memory perception and sense of ownership over the outputs
Despite being a relatively small study the findings were clear The group working with ChatGPT showed the lowest brain activity among the three mainly in areas related to creative thinking working memory and cognitive control.
When writing without aids participants activated many brain networks and created connections between different areas an activity indicating deep cognitive engagement.
But when relying on artificial intelligence this engagement weakened significantly and thinking became shallow based mainly on input-output.
The researchers called this phenomenon cognitive debt the more we rely on the tool the less our brain works and when asked to think alone it is difficult to return to previous performance level
Moreover when participants were asked to recall sentences they wrote with the language model they struggled to do so and even reported a sense of alienation from the outputs.
In other words not only did the brain work less the sense of control over writing was also impaired
Does this mean we should give up on artificial intelligence Not necessarily Just like calculating with a calculator or navigating with GPS software ChatGPT and similar tools can be excellent aids as long as we know when and how to use them.
The key is awareness not turning the tool into a substitute for thinking but a lever that expands it. Letting the brain work even when effort could be saved.
Combining ideas from the model with personal connections memory and human reasoning. This way we can enjoy the capabilities of artificial intelligence without losing our natural brain function