After Elon Musk, Facebook wants to connect brains to machines
A team of US researchers funded by Facebook has been able to translate brain activity into words via algorithms, one step closer to the ambition shared by other companies, such as Elon Musk’s create a direct link between the human brain and machines.
The scientists are UCSF Californian university have this week released a study showing their progress in creating a brain-computer interface: the activity of neurons is transmitted to the machine thanks to the implants, and decoded by algorithms, in a specific context (a limited choice of questions and answers).
Ultimately, the ambition would be to use a less restrictive method than implants, such as a pair of augmented reality glasses, equipped with sensors and controlled by thought.
“In the next ten years, the ability to type directly from our brain may be considered normal,” said Facebook Tuesday in an online article reviewing the project launched two years ago. “Not so long ago, it was science fiction. Now this prospect seems attainable. ”
Elon Musk, the whimsical boss of Tesla and Space X, struck the spirits 10 days ago by claiming that start-up Neuralink had achieved an experiment in which a monkey was controlling a computer directly from his brain.
The company is expected to begin testing on humans as early as 2020, in order to fight certain diseases affecting the brain or spinal cord.
A health goal that also guides the UCSF team’s research: to make people speechless because of paralysis, spinal cord injuries or neurodegenerative diseases.
“At this point, paralyzed patients with speech loss can only use eye movement or muscle contraction technologies to spell words very slowly on a screen,” says Eddie Chang, a neuroscientist at the University of Toronto. California University.
“Yet, in many cases, the ability to express themselves is always present in their brains. We just need the technology to enable them to use it smoothly. ”
The study published in the journal Nature Communications details how the scientists managed to train the machine to translate the signals received through the implants, while the volunteers spoke aloud. The artificial intelligence system was guided by a context of multiple choice questions posed to people.
The “Steno project” is studying the possibility of making a connected accessory to tap on a screen simply by imagining talking in his head. It is funded by a technology giant (Facebook Reality Labs) lab doing research on augmented and virtual reality technologies.
“Our progress shows what might one day be like interacting with smart glasses,” tweeted Andrew Bosworth, vice president of Facebook’s consumer equipment division.
However, many neurology experts have expressed doubts about the real possibilities of these technologies, even in the medium term, given how we mobilize multiple parts of the brain, even for simple tasks.
This intimate relationship of which Elon Musk speaks is “rather a vision of a very distant future,” said Andrew Hires, assistant professor of neurobiology at the University of Southern California. “It is not certain that we will arrive at this stage one day.”