If he does not want to end up as a vulgar slave of artificial intelligence, humans have no choice: he will eventually have to merge with it. So let’s connect our brains to computers, thanks to implants placed in the cortex, and our memory, calculation and reasoning capacities will finally be worthy of an algorithm! This is the marvelous program that presided over the creation, in 2016, of the Californian start-up Neuralink, revealing the transhumanist ambitions of its boss Elon Musk.
Seven years later, the indefatigable billionaire says his brain implant is about to be tested on human subjects, possibly before the summer of 2023. For now, the United States Medicines Agency (FDA) has not yet given its approval. But scientists specializing in the field are taking this new player seriously.
“Musk comes late, but he left very quickly, thanks to colossal investments. As a result, he surrounded himself with the best and developed an implant that seems to perform very well, even if the stability over time of its performance is not yet known. summarizes Blaise Yvert, research director at Inserm. According to him, the new Neuralink implant could prove to be more precise than those that already exist, due to unprecedented flexibility for use in humans and a very large number of electrodes.
Restoring movement to quadriplegic patients
While it certainly gets the most media coverage, Neuralink is not the only laboratory to develop brain-machine interfaces (BCIs) – France has been at the forefront in this field, behind the United States, for a twenty years. In these two countries, as well as in the Netherlands in particular, a handful of people have already been equipped with such devices in the context of highly supervised clinical trials. Most of them are quadriplegic patients who thus learn to remotely direct the movements of a robotic arm, a wheelchair… even a four-limbed exoskeleton, as was the case in 2019 at the Clinatec laboratory. from the CEA, for a young man from Lyon who was disabled following an accident.
“There’s nothing magical about it,” immediately rectifies Blaise Yvert, concerned to read here or there that implanted people can control machines “by thought”. “The brain implant does not collect the thought, but the electrical activity of a limited area of the brain when the patient imagines, for example, making a movement or pronouncing words”, he says. Indeed, when you talk to yourself in your head, you activate motor neurons linked to the articulation of language…
Once brain activity is recorded in this area of the motor cortex, it is processed by a computer system which then translates it into movement – or language – using mathematical algorithms. Such feats cannot take place without real cooperation from the patient: the patient must, in order to “train” the algorithm, imagine many times making such a gesture or pronouncing such a word.
More or less invasive devices
“Today, the vast majority of teams specializing in ICM research are working on non-invasive systems, with electrodes placed on the surface of the scalp (EEG helmet)”, explains Guillaume Charvet, head of medical devices at the CEA’s Clinatec laboratory. Only 10% are developing implants to be placed under the skull – either on the surface of the motor cortex (semi-invasive) or intracerebrally (invasive).
Semi-invasive implants such as those being developed by the CEA can capture the signals of an entire population of neurons in one fell swoop. Above all, they offer a stable signal over time, which suggests the possibility of home use, with patients implanted for several years. But these systems are less accurate than invasive implants, as already exist in the United States, for measuring the activity of single neurons.
The Neuralink implant is invasive, and has a number of electrodes that put the competition to shame: 1,024, compared to a hundred for the Utah Array system, the only intracortical implant tested on humans to date ( twenty years already). Elon Musk’s device will also be more flexible, while the Utah Array is presented as a rigid square with a side of 4 millimeters. However, its longevity remains unknown.
Send SMS faster
While research on ICMs has long focused on restoring gross motor skills, it is focusing on increasingly dexterous and complex gestures, such as handwriting. In May 2021, a team from Stanford University published in the journal Nature the results of an experiment on a 65-year-old implanted man with a paralyzed hand. The latter was able to communicate about 90 characters per minute – a typing speed comparable to those of individuals of the same age typing on their smartphone (115 characters per minute)!
Precisely, among the advantages of having an implant, according to Elon Musk, there would be the possibility of sending text messages directly from his brain, without “wasting time” typing them on his phone… “This leaves one wondering about the interest of such a device! » reacts Jérémie Mattout, from the Lyon Neuroscience Research Center (Inserm). “Note, however, the evolution of these announcements. We are far from the first completely fantastic promises of soon being able to read our neurons and download our memory onto a computer’s hard drive. »
The boss of Neuralink would he have come up against the wall of reality? Be that as it may, the question of “recreational” use of brain implants is far from unanimous. The scientists interviewed all insist on the risk-benefit balance, the danger being that the chip and the brain are not biocompatible. Last December, the Reuters agency revealed that 1,500 animals had died as part of trials for the future Neuralink implant. A federal investigation has been opened in the United States for violation of animal welfare laws.
The concern of free will
Beyond Elon Musk’s unique approach, the very principle of the brain implant is not without raising many ethical questions. “The first is whether an implanted human is still a human”begins Eric Fourneret, philosopher and lecturer at the Catholic Institute of Lille, researcher in the international project “For a new humanism in the age of neuroscience and artificial intelligence” (1). “There is a form of hybridization there, with two entities cohabiting, what is more in the brain, the place where our consciousness. »
Just as our smartphones use intuitive writing when typing a message, won’t artificial intelligence also be able, in the long term, to make suggestions to humans, to which it would be incorporated through a brain implant? What place then for free will? “When we speak, we can control what we say, adds Eric Fourneret. Would this still be the case with a device that externalizes everything we say to ourselves in our heads? » Concerns are already being heard around the confidentiality of data from an implanted brain.
According to neuroscientists, however, a “trivialization” of ICM seems difficult to envisage in the medium term. “Only a few experiments have been done so far”recalls Jérémie Mattout, who suggests a comparison with the conquest of space. “Just as we made a handful of people walk on the Moon, we demonstrated the technological feasibility of ICMs. This does not make them secure or accessible devices. » Not to mention the exorbitant cost of these implants which are still at the prototype stage.
———
France as a pioneer in this area
Invasive and semi-invasive brain-machine interfaces (BCIs), still in the research stage, have been tested on very few patients around the world. This technology took off at the turn of the 20th century, and is now developing in the United States, France, the Netherlands, Germany and China.
Other types of brain implants already have a solid and widespread clinical application, such as deep brain stimulation (DBS) by placing a probe composed of electrodes in the brain. It is used in 5 to 10% of Parkinson’s patients to limit tremors.
This type of implant is not always considered a brain-machine interface, because it is not based on the recording of brain activity but on the administration of small pulses of electrical current.
Neurosurgeon Alim-Louis Benabid was a pioneer of DBS in the 1980s. In 2011, he founded the Clinatec laboratory, which today develops a semi-invasive ICM. The Neuralink implant is invasive.