Here’s how the technology behind ChatGPT could make reading minds a reality

() — On a recent Sunday morning, I found myself wearing an ill-fitting pair of scrubs, lying on my back in the claustrophobic confines of an fMRI machine at a research facility in Austin, Texas. “The things I do for television”, I thought.

Anyone who has had an MRI or fMRI will tell you how noisy it is: electrical currents swirl around creating a powerful magnetic field that produces detailed scans of your brain. This time, however, I could barely hear the loud start of the mechanical magnets, I was given a pair of specialized headphones that started playing segments from the “Wizard of Oz” audiobook.

Because?

Neuroscientists at the University of Texas at Austin have discovered a way to translate scans of brain activity into words using the same artificial intelligence technology that powers the innovative ChatGPT chatbot.

The breakthrough could revolutionize the way people who have lost the ability to speak can communicate. It’s just a pioneering AI application developed in the past few months as technology continues to advance and seems poised to touch every part of our lives and society.

“Well, we don’t like to use the term mind reading,” Alexander Huth, an assistant professor of neuroscience and computer science at the University of Texas at Austin, told me. “We think it evokes things that we’re not really capable of.”

Huth volunteered to be a research subject for this study, spending more than 20 hours in the confines of an fMRI machine. [resonancia magnética funcional] listening to audio clips as the machine took detailed images of his brain.

An artificial intelligence model analyzed your brain and the audio you were listening to, and over time, was eventually able to predict the words you were hearing simply by looking at your brain.

The researchers used San Francisco-based startup OpenAI’s first language model, GPT-1, which was developed using a massive database of books and websites. By analyzing all this data, the model learned how sentences are constructed, essentially how humans speak and think.

The researchers trained the AI ​​to analyze the activity in the brains of Huth and other volunteers as they listened to specific words. Eventually, the AI ​​learned enough to be able to predict what Huth and others were hearing or seeing simply by monitoring their brain activity.

I spent less than half an hour in the machine and, predictably, the AI ​​couldn’t figure out that I had been listening to a portion of the “Wizard of Oz” audiobook that described Dorothy walking down the yellow brick road.

Huth listened to the same audio, but because the AI ​​model had been trained on his brain, it was able to accurately predict parts of the audio he was listening to.

While the technology is still in its infancy and shows great promise, the limitations may be a source of relief for some. AI can’t easily read our minds, yet.

“The real potential application of this is to help people who can’t communicate,” Huth explained.

He and other UT Austin researchers believe the innovative technology could be used in the future by people with “locked-in” syndrome, stroke victims and others whose brains work but cannot speak.

“Ours is the first demonstration that we can get this level of precision without brain surgery. So we think this is the first step on this journey to really help people who can’t speak without neurosurgery,” he said.

While revolutionary medical advances are certainly good news and can be life-changing for patients battling debilitating ailments, it also raises questions about how the technology could be applied in controversial settings.

Read minds

Could it be used to extract a confession from a prisoner? Or to expose our deepest, darkest secrets?

The short answer, Huth and his colleagues say, is no, not at the moment.

For starters, brain scans must be performed in an fMRI machine, artificial intelligence technology must be trained on an individual’s brain for many hours, and, according to the Texas researchers, subjects must give consent. If a person actively resists listening to the audio or thinks about something else, the brain scans will not be successful.

“We believe that everyone’s brain data should be kept private,” said Jerry Tang, the lead author of a paper published earlier this month detailing his team’s findings. “Our brains are like one of the last frontiers of our privacy.”

Tang explained: “Obviously, there are concerns that brain scrambling technology could be used in dangerous ways.” Brain decoding is the term that researchers prefer to use instead of mind reading.

“I feel like mind reading evokes this idea of ​​getting to the little thoughts that you don’t want to let go, little reactions to things. And I don’t think there’s any suggestion that we can actually do that with this kind of approach,” Huth explained. “What we can get is the great ideas that you are thinking of. The story that someone is telling you, if you’re trying to tell a story inside your head, we can come up with that too.”

Last week, the creators of the generative AI systems, including OpenAI CEO Sam Altman, descended on the US Capitol to testify before a Senate committee about lawmakers’ concerns about the risks posed by the powerful technology. Altman warned that AI development without guardrails could “cause significant harm to the world” and urged lawmakers to implement regulations to address the concerns.

Echoing the AI ​​warning, Tang told that lawmakers need to take “mental privacy” seriously to protect “brain data,” our thoughts, two of the most dystopian terms I’ve heard in the age of the AI.

While the technology at the moment only works in very limited cases, that may not always be the case.

“It is important not to have a false sense of security and think that things will be like this forever,” Tang warned. “Technology can get better and that could change how well we can decode and change if the decoders require the cooperation of a person.”