Neuroscientists Can Now Read People’s Thoughts Using Brain Scans and AI

In a groundbreaking new study in Nature Neuroscience, scientists have demonstrated the ability to interpret human thoughts by parsing data from functional magnetic resonance imaging (fMRI) scans with artificial intelligence.

This avenue of research is already raising concerns for many as it is tantamount to mind-reading. However, there are a few important caveats that make it a bit less intimidating. The study says that successful decoding of thoughts requires cooperation from the subject, both to train the model and to use it after it’s trained.

This method is also extremely expensive and complex as it requires fMRI machines, which can cost $1-3 Million, as well as complex AI models and a ton of time and data.

The paper, authored by Jerry Tang, Amanda LeBel, Shailee Jain and Alexander G. Huth, is not definitive proof of mind-reading by itself. Because the paper is so new, it has yet to be replicated, which is a necessary step in the scientific process. It also only involved 3 test subjects.

How Does It Work?

The process began with having the test subject lie in an fMRI machine for at least 16 hours each. Then they listened to stories from a podcast while the fMRI closely studied the blood flow in their brain, collecting as much data on the stimuli as possible.

With this extensive data, computational neuroscientists Jerry Tang and Alexander Huth matched patterns of brain activity to certain words and ideas in the stories. They used a complex large language model, based on GPT, to help decode the data.

With a general idea of how the brain blood flow matched with words, the researchers were able to start predicting undecoded thoughts. However, the model didn’t solely rely on brain activity. It first predicted which words would be the most likely to come next and then used the brain activity data to choose which one to pick.

The model seems to struggle with choosing the exact correct word, but its ability to predict the correct general idea of the thought was impressive. Huth gave the example of it predicting the sentence “She has not even started to learn to drive yet,” when the subject thought “I don’t have my driver’s license yet.”

What’s the Point of this Research?

There are more potential uses for this kind of technology than listening into people’s thoughts. For one, it could lead the scientific community to a much greater understanding of the brain. This kind of foundational research can be vital catalysts to exponential growth of understanding in their fields.

One of the most promising applications of this research is its potential to assist individuals with communication difficulties, such as those suffering from locked-in syndrome or severe paralysis.

By decoding their thoughts through brain scans, the AI could potentially enable these individuals to communicate their needs and emotions, dramatically improving their quality of life.

While the potential of this research is exciting, it is essential to consider ethical concerns and potential risks as well. Issues surrounding privacy and consent need to be carefully evaluated and addressed to ensure responsible development.

Related Articles:

• Why the Warnings of Airport Juice Jacking Are Overblown at Best
• Best Meme Coins to Buy in 2023
• India’s Most Valuable Startup Byju’s Raided Under Anti-Money Laundering Law