Computers have been able to read human thoughts almost in real time

By combining implantable devices in the brain with a special software, scientists were able to use computers to "read" the mind of a person almost as soon as they thought, from that point forward. Guess what content they are about to say with 96% accuracy.

The human brain is able to look at a 2-dimensional image on a paper or computer screen, then transfer this image to something inside the human brain in an extremely fast way. However, this neurological process has so far been a mystery to scientists. In order to clarify that, at the same time seek ways to help computers to collect and predict human thinking, a group of scientists led by neuroscientist Rajesh Rao and neurosurgeon Jeff Ojermann at the University Washington has used computers and implant devices to decode brain signals at almost the same rate as human consciousness.

The human brain is capable of looking at a 2-dimensional image on paper or computer screen.

In the study, they asked for the support of seven patients who were treated for epilepsy. Drug therapy has no effect on reducing seizures and these patients have temporarily implanted electrodes into the brain to determine, stimulating the focal points when seizures occur. And according to the research team, this is an opportunity for them to conduct the tests.

After setting up the experiment, patients will be given a random view of a series of images of human faces, houses, fully gray images displayed on a computer screen at a rate of 400 milliseconds. . Their mission is to find a photo with the house upside down.

In the meantime, the electrodes in their brains will be connected to a software capable of extracting two distinct properties of the signal in the brain as "potential linkage between events" (when an amount is Large neurons simultaneously light up to meet an image) and change of "broadband spectrum" (those signals are maintained after viewing an image).

As soon as the photo flashes on the screen, the computer will sample and digitize the signal coming from the brain at a rate of 1000 times per second. This detail allows the software to determine the combination of the electrode position and the signal that best corresponds to what the patient sees. Professor Rao said: "We have seen very different responses from different brain regions, some are sensitive to face images and others are sensitive to the home image."

Transplant devices to decode brain signals at almost the same rate as human consciousness.

After completing the software preparation, the team showed the patients a whole new set of photos. Although computers were previously "taught" the reaction of people to old photos, but for new photos (also the subject is face, house, gray but other photos), it also can predict the brain region will be stimulated with accuracy up to 96%. Interestingly, this result is returned at almost the same rate as human cognitive speed.

However, because only certain objects are taught, the computer in the study can only identify the potential for linking between potential events and changes in the broadband spectrum when users meet topics. above. Therefore, in order for it to understand many other people's perceptions, it is necessary to use an artificial intelligence with a large artificial neural network to have enough databases and computing power.

Although the results in this study are very promising, there are still certain limitations. In addition to database constraints, this study is based on owners who have large differences, while whether between humans and animals, or even faces of different people, is expected. It is unclear whether or not to guess. However, this type of technology is really attractive, contributing to the smartness of AI .. one level, and the concern that AI will counteract people like in the Terminator movie. increase.