This past summer, researchers at UC San Francisco announced in Nature Communications that they successfully decoded a small set of words and phrases directly to text from brain activity. This brain-computer interface is the result of researchers studying brain mapping teaming up with Facebook for develop AI that can actually read minds verbatim.
Now that a couple words have been decoded, the goal is to hit the 1,000-word vocabulary mark. While the goal is to develop a wearable device that allows users to think words onto a screen much like voice-to-text tech we have today, Facebook reports that the current technology is too slow, bulky, and requires a brain implant. The study subjects that produced successful thought to text interface were three epileptic patients who were being treated at the medical center at UC San Francisco. All three had electrodes implanted in their brains already to record brain activity to locate the source of their seizures. Their hope is that they will be able to decode words using this brain imaging to text technology in real time with hardware that is non-invasive and relatively quick, working at a target rate of 100 words per minute with less than 17% error rate.
Before you take out your tin foil hat, like human microchipping, there is more to this new AI development than digital autocratic agendas. Similar to the upside of human microchipping is that through brain to text word decoding, individuals who have lost their speech and movement abilities due to strokes, paralysis, and other severe medical conditions will be able to communicate again. We may even be able to decode the thoughts of people in a coma. Of course, the downside opens an equally astronomical array of possibilities.
While this technology still has a long way to go before we should start being concerned, elated, or both, successfully interpreting words from brain scans in real time is a first achievement in its field, and signals another monumental shift in the way we communicate may be on the horizon.
