A team of Japanese scientists is one step closer to building computer systems that can actually think for themselves — in a sense, at least.
Researchers at Kyoto University have found success using machine learning to literally read minds. There’s no psychic abilities here, though — just computers with excellent artificial intelligence.
The key here is the ability to decode thoughts.
To do this, scientists need to be able to translate brain scans to visual thoughts. It’s not an easy task, but their platform, BioRxiv, is becoming more and more capable.
It’s a big step.
In the past, A.I. could detect simple visualizations from a brain scan (like a shape or a letter), but the team from Kyoto has a system that can detect more complex images, with multiple colors and structures.
Basically, the A.I. can detect all sorts of multiple levels.
Rather than decoding thoughts on a pixel-by-pixel basis, the key to the breakthrough made by the new A.I. lies in its ability to recognize shapes, not just small components of pixels.
Study participants were shown various images.
These images include stuff from nature, geometric shapes, and letters. Brain scans were conducted while participants visualized the shapes, and the A.I. recorded the person’s thoughts. The results, needless to say, were pretty stunning.
The A.I. doesn’t quite work in real time.
After the brain scans are conducted, the A.I. needs to decode the information, processing it into an image that represents what was on the person’s brain.
For the images used, the results were pretty much bang-on.
When an image was fresh on the person’s mind, the picture recorded from the brain scan matched up with the picture used almost perfectly.
The tech is also studying people’s brains for other memories.
The technology isn’t quite there yet, but it’s getting better and better at capturing random memories, not just the visualizations included in the study.
It’s an exciting new frontier in artificial intelligence.
A.I. is a new frontier, being worked on from everybody from the Kyoto researchers to entrepreneurs to Google. It’ll be crazy to see where we’ll be a few years from now.