At the end of August, Neuralink, a neurotechnology company founded by Elon Musk, demonstrated its latest implantable brain chips that could in the future lead to powerful brain–machine interfaces. In Musk’s own words: “It’s kind of like a Fitbit in your skull, with tiny wires.” The technology enables precise recordings of brain activity in real time as well as brain stimulation. The list of future applications is long and ranges from curing diabetes and restoring movement in paralyzed patients through to playing computer games and driving cars via the power of thought.
As a former neuroscientist and currently a machine learning engineer at AMPLYFI, I have been following Neuralink since its beginnings and was excited to hear about the latest developments. During my PhD, I carried out experiments to understand the role of brain oscillations in information processing; now at AMPLYFI I apply my knowledge and experience to build artificial neural networks to extract insights from millions of documents published on the internet every day.
In the live-stream demo, Musk introduced a pig called Gertrude who had been fitted with a tiny chip that recorded the animal’s neural activity and sent it wirelessly to a screen and speakers in real time. To a neuroscientist, the scientific aspect of the demo revealed nothing new; every day researchers around the world record brain activity from implanted electrodes and perform deep brain stimulation. Although other labs demonstrated that a monkey can control a robotic arm purely with its thoughts back in the late 2000s, Musk didn’t show how his team could practically utilize the signals recorded in real time. Even the Neuralink’s electrode density (1,024 channels) is not especially impressive these days.
So, is Neuralink’s technology over-hyped? No, it is not. Although Musk didn’t show anything that hasn’t been done before in terms of science, there are a number of important engineering advances that should be highlighted. The coin-sized chip itself is a piece of cutting-edge engineering and the electronics that relay neural signals wirelessly from the brain is state-of-the-art. The chip is charged with an induction coil and can be easily replaced as technology improves. The robotic implantation technology was another highlight. The updated prototype of the sewing robot, encased within a smooth white plastic helmet, will be able to carve out a circle of bone and thread electronics into their brains in a procedure that should take under an hour while causing no lasting damage to the brain. “I could have a Neuralink right now and you wouldn’t know it,” Musk said.
Despite the promising advances in engineering, there are some fundamental challenges that Neuralink’s team will have to overcome if they are to deliver on their ambitious promises. Understanding how brain activity underpins cognition and information processing is extremely difficult. Although there has been progress in decoding motor function, speech, visual, and auditory perception, much more research is needed to develop reliable systems that can translate brain activity into the multitude of human cognitive and behavioural processes. And perhaps the most important question is, once the technology matures, will people be comfortable being implanted with devices that could potentially be hacked and make us buy certain products, fall in love, influence the way we vote, or even go to war?