Scientists have achieved a new milestone in Artificial Intelligence (AI) technology by developing a mind-reading program that can transcribe the gist of what people are thinking using brain scans. While this advancement is seen as an encouraging step to helping those who have lost their speech function, it also raises ethical concerns regarding the protection of mental privacy in an AI-characterized future.
The study, recently published in Nature Neuroscience, viewed the brain activity of three people as they listened to spoken narrative stories or even formulated their own tales in their minds. Through mapping the areas of the brain that process language, the researchers were able to connect this data with a neural network language model that used GPT-1, the former AI system for ChatGPT. After being trained to predict how each individual brain responds to perceived speech, the decoder was then able to pinpoint the closest option to accurately reconstruct continuous language.
The authors of the study, Jerry Tang and Alexander Huth from the University of Texas at Austin, note that this system did not require an invasive brain implant and was able to pick out the “gist” of what the participants were thinking. They add that the model did stumble on certain words and personal pronouns, though its accuracy did not decrease when participants made their own stories in their minds or watched silent movies.
In view of this progress, bio-ethics professor, David Rodriguez-Arias Vailhen from Granada University, states that this new machine system is bringing us closer to a possible future where machines can read minds and transcribe our thoughts, even against our will. With that, he advocates that legislative rules should be set in place to protect mental privacy. On the contrary, the authors state that participants were still able to easily avoid the decoder’s detection by either counting by 7s or by naming and imagining animals.
In the pursuit of further developments, the researchers want to increase the speed of the system to be able to decode the brain scans in real-time. Ultimately, in an AI-dominated world, measures to protect the privacy of our thoughts must be taken into account, as it is the key to continued free thinking.
