[imagesource:maxpixel]

Yes, it is as terrifying as it sounds. Better grab your tinfoil hat, you are going to need it for sure this time.

Scientists have released a breakthrough study into using an invented language decoder that can translate a person’s thoughts into text, using an artificial intelligence (AI) transformer similar to ChatGPT.

The study reveals the ‘first time that continuous language has been non-invasively reconstructed from human brain activities’. In other words, the decoder is able to read your thoughts based on the data from an MRI machine.

Sounds scary? It is, but luckily it’s not that simple, so SAPS won’t be getting their hands on this tech yet, or have anyone that can work the many buttons.

The decoder was able to interpret stories people were watching, listening to, or even just thinking of, based on the data from an MRI machine. Although not 100% accurate, the decoder ‘allows it to read peoples’ minds with unprecedented efficacy’.

Currently language decoding can be done via implanted devices that require invasive neurosurgery, but with this decoder, scientists hope to make the process ‘less intrusive’. As if reading your mind is anything but intrusive.

“The goal of language-decoding is to take recordings of a user’s brain activity and predict the words that the user was hearing or saying or imagining. Eventually, we hope that this technology can help people who have lost the ability to speak due to injuries like strokes, or diseases like ALS.”

Vice quotes Jerry Tang, a graduate student in computer science at the University of Texas at Austin, who led the study, as saying they were able to achieve the results by having three volunteers spend 16 hours listening to a series of stories in an MRI machine.

The decoder then used ‘Reddit comments and autobiographical stories in order to link the semantic features of the recorded stories with the neural activity captured in the fMRI data’. This allowed them to associate words and phrases with certain brain patterns.

For the second part of the test, the volunteers listened to different stories, and using the data from the first test, the decoder was able to ‘read’ their brain activity, and translate it into text. The decoder is however not perfect yet and changes the semantics of certain sentences, but the results are both incredible and terrifying.

For instance, the sentence “I don’t have my driver’s license yet” was decoded to “She has not even started to learn to drive yet.”  Pretty damn close if you ask me.

“What we get out is not the exact words that somebody heard or spoke, it’s the gist. It’s the same idea but expressed in different words.”

The potential of this tech is enormous, but thankfully the inventors are not oblivious to the potential dangers should this be used for nefarious reasons.

In the end, mental privacy is probably the last bit of real privacy we have.

“For these and other unforeseen reasons, it is critical to raise awareness of the risks of brain decoding technology and enact policies that protect each person’s mental privacy.”

Nice sentiment, but all the policies in the world will not be able to protect us from the ‘thought police’ when they come knocking.

It feels as if we are playing around with technology that no one is prepared for, and the consequences could be dire.

[source:vice]