University of California's New Mind-Reading Device Can Translate Brainwaves Into Words
It's strange to think that the first mainstream voice assistant, Siri, was only introduced in 2011. Now, seven years later, researchers from the University of California have created a device capable of translating brainwaves into text with 90 percent accuracy.
Other companies, like the Boston-based startup Neurable, have been able to harness brainwaves to navigate menus or create effects in VR, but the prospect of a mind-reading brain-to-text device may cross the line between technological breakthrough and privacy nightmare.
A new paper, published in the Journal of Neuroengineering, outlines the results of the initial experiments and follows on from the team's previous research in 2016 into neural speech recognition (NSR).
The system works by implanting electrodes over the brain's surface and sensing brain activity, including signals related to combinations of consonants and vowels. After processing these signals, the device is able to display sentences picked up from the subject's brain in real time.
Especially impressive is the device's ability to process words it has not seen before.
According to David Moses, the lead author of the newly published study, "No published work has demonstrated real-time classiﬁcation of sentences from neural signals. Given the performance exhibited by [the machine] in this work and its capacity for expansion, we are conﬁdent in its ability to serve as a platform for the proposed speech prosthetic device."
As Moses mentions, the device's primary goal is to help those who have lost the ability to speak. However, it doesn't take much imagination to think of other potential uses for the device, such as neural prosthetics to allow anyone to send messages to each other without touching their phones or keyboards.
One of the inevitable questions surrounding this technology is going to be how to keep unwanted thoughts from contaminating an intended message.
Another key issue is going to be privacy—when thoughts can be transcribed digitally in real time, the prospect of hacking a device to tap into the stream of messages could turn neural speech recognition into a window to our private thoughts.
We want to say this technology is heralding the beginning of a brave, new world, but that may not be the right dystopia for a device like this—it sounds closer to 1984.