Listen to the Podcast:
Artificial intelligence (AI) is currently being used by scientists to scan human brain waves and translate them into text.
A study in the journal Nature Neuroscience claims that the AI models underlying Google’s Bard and Open AI’s ChatGPT can read someone’s mind and mimic their brain activity in a stream of text.
Jerry Tang, a computer science PhD candidate, and Alex Huth, an assistant professor of neurology and computer science at UT Austin, served as the study’s principal investigators.
Three people were asked to listen to stories as part of the study. Then, the people were put in MRI machines. Scientists say they were able to get the text of the people’s thoughts without putting anything in their brains.
The results of the study show that non-invasive language brain–computer connections are possible. The study says, “These kinds of systems could be especially helpful for people who can’t physically speak, like those who have had a stroke, and help them communicate better.”
At the time, the semantic decoder can only get the “gist” of what someone is thinking. Also, only about half of the time did the AI decoder come up with a text that was close to what the person was thinking.
The study gave an example of how the decoder worked by showing how a test subject heard the phrase “…I didn’t know whether to scream, cry, or run away. Instead, I said, “Leave me alone, I don’t need your help,” and Adam then disappeared.”
The decoder turned this part of a phrase into “…started to scream and cry, and then she just said, “I told you to leave me alone. You can’t hurt me anymore. I’m sorry,” and he stormed off.
“Compared to what’s been done before, which was usually single words or short sentences, this is a big step forward for a noninvasive way. In a study posted on the UT Texas website, Professor Huth says, “We’re getting the model to decode continuous language for long periods of time with complex ideas.”
The experts are also worried about the privacy of people‘s thoughts. Jerry Tang said, “We are very worried that it could be used for bad things, and we have worked to keep that from happening.” “We want to make sure that people only use these types of technologies when they want to and when it helps them.”
