Neuralink has demonstrated a prototype of its brain-machine interface that currently works in pigs.
At a Friday event, Elon Musk revealed more details about his mysterious neuroscience company Neuralink and its plans to connect computers to human brains. While the development of this futuristic-sounding tech is still in its early stages, the presentation was expected to demonstrate the second version of a small, robotic device that inserts tiny electrode threads through the skull and into the brain. Musk said ahead of the event he would “show neurons firing in real-time. The matrix in the matrix.”
And he did just that. At the event, Musk showed off several pigs that had prototypes of the neural links implanted in their head, and machinery that was tracking those pigs’ brain activity in real time. The billionaire also announced the Food and Drug Administration had awarded the company a breakthrough device authorization, which can help expedite research on a medical device.
Like building underground car tunnels and sending private rockets to Mars, this Musk-backed endeavor is incredibly ambitious, but Neuralink builds on years of research into brain-machine interfaces. A brain-machine interface is technology that allows for a device, like a computer, to interact and communicate with a brain. Neuralink, in particular, aims to build an incredibly powerful brain-machine interface, a device with the power to handle lots of data, that can be inserted in a relatively simple surgery. Its short-term goal is to build a device that can help people with specific health conditions.
The actual status of Neuralink’s research has been somewhat murky, and Friday’s big announcement happened as ex-employees complain of internal chaos at the company. Musk has already said the project allowed a monkey to control a computer device with its mind, and as the New York Times reported in 2019, Neuralink had demonstrated a system with 1,500 electrodes connected to a lab rat. Since then, Musk has hinted at the company’s progress (at times on Twitter), though those involved have generally been close-lipped about the status of the research.
Musk opened Friday’s event by emphasizing the wide variety of spinal and neurological conditions — including seizures, paralysis, brain damage, and depression — that Neuralink technology could help treat.
“These can all be solved with an implantable neural link,” said Musk. “The neurons are like wiring, and you kind of need an electronic thing to solve an electronic problem.”
But it’s worth highlighting that Musk wants Neuralink to do far more than treat specific health conditions. He sees the technology as an opportunity to build a widely available brain-computer interface for consumers, which he thinks could help humans keep pace with increasingly powerful artificial intelligence.
So while modest, Neuralink’s research already foreshadows how this technology could one day change life as we know it. At the same time, it’s a reminder that the potential, eventual merging of humans with computers is destined to introduce a wide range of ethical and social questions that we should probably start thinking about now.
Neuralink wants to link your brain with computers, but that will take a while
Founded in 2016, Neuralink is a neuroscience technology company focused on building systems with super-thin threads that carry electrodes. When implanted into a brain, these threads would form a high-capacity channel for a computer to communicate with the brain, a system supposed to be much more powerful than the existing brain-machine interfaces being researched.
One major barrier to inserting these incredibly tiny wires, which are thinner than a strand of human hair, is actually getting them past the skull and into the brain. That’s why Neuralink is also developing an incredibly small robot that connects the electrode to humans through surgery that’s about as intensive as a Lasik eye procedure. On Friday, Musk outlined how the company hopes to do the procedure without general anesthesia in a single-day hospital stay. That’s the goal at least, and would represent a huge leap forward from previous brain-machine interfaces, which have required more invasive surgeries.
“We’ve been connecting forms of computers to brains for 20 or 30 years already,” Nolan Williams, the director of Stanford’s Brain Stimulation Lab, told Recode, referencing deep stimulation used for patients with Parkinson’s as one example of connecting a brain and a computer.
“The brain itself uses certain frequencies and certain kinds of electrical thresholding to communicate with itself,” Williams explained. “Your brain is a series of circuits that kind of intercommunicate and communicate between themselves.”
A screenshot of the demo shows how a prototype can track the neural spikes of a pig that’s had the device implanted.
Essentially, a brain-machine interface can use the electricity the brain already uses to function along with a series of electrodes to connect the brain with a machine. Neuralink cites previous examples in which humans have used electrodes to control cursors and robotic limbs with their minds as the basis for its system. But what’s novel about Neuralink’s plan is making the process of connecting a device with the brain minimal, while also massively increasing the number of electrodes engaged. The company wants to make brain-machine interfaces not only easier to install but also more powerful.
The Neuralink surgical robot, revealed on Friday, is supposed to manipulate and insert the tiny threads into the brain.
As the focus of Friday’s event, Musk showed what the second generation of that robot will look like: a large white structure with five degrees of freedom.
“The robot is a super complicated, highly-precise machine which is able to both capture your brain and then with almost a sewing machine-like, micro-precise needle and thread, place the neural threads in the exact right location based on the surgeon decisions around what the safe locations are for the threads to be inserted,” Afshin Mehin, a designer and founder of the firm Woke, which worked on the robot’s outer device that holds the needle, told Recode.
The machine operates at a very small scale, and Neuralink hopes to expand its capabilities. For instance, the current robot has a 150 micrometer gripper, and an even tinier needle — less than 40 micrometers — which can “grasp the implant’s threads then precisely insert each into the cortex while avoiding visible vasculature,” according to Neuralink robotics director Ian O’Hara. He added in an emailed statement that, while the robot currently handles only the insertion of the threads, Neuralink is working to expand the robot’s role in surgery to increase the number of patients it can help and make the procedure shorter.
Musk said that, in the past year, Neuralink simplified its plans for a wearable device that connects to the threads implanted in the user’s brain. While the first generation of this device would have been installed behind a person’s ear, the newest version is a small, coin-size device that would sit under the top of their skull.
“It’s kind of like a Fitbit in your skull with tiny wires,” explained Musk, who compared the device to a smart watch.
The research is still in early stages and, as it advances, will likely require focusing on how the technology can help people with specific, severe health conditions first, according to Stanford neurosurgery professor Maheen Adamson. While the medical applications of such technology could be wide-ranging, moving it from its current, nascent state will require the close oversight of the Food and Drug Administration, which would not comment specifically on Neuralink.
Why Elon Musk fears artificial intelligence
Again, Neuralink’s ultimate plans will go beyond treating specific conditions. The company has said it hopes to allow people to “preserve and enhance” their brains and to “create a well-aligned future.” While that might not sound like a particularly pressing need to the average person, the project fits into Musk’s long-standing concerns about artificial intelligence. Musk has previously said the technology could be more dangerous than nuclear weapons, and warned that AI could become too powerful, too quickly, preventing humans from keeping it in check.
The ultimate goal for Neuralink, Musk explained at a 2019 launch event, is a “full brain-machine interface” that will achieve a “symbiosis with artificial intelligence.” But again, that is still far off.
Brain-machine interfaces are nothing new, but they raise ethical concerns
Neuralink and Musk are not the only ones interested in brain-machine interfaces. Facebook, for instance, is hard at work on its own brain-machine interface research with the University of California San Francisco. The company has its eye on creating a “hands-free” way of communicating with computers and has shared some very preliminary results. Last year, Facebook purchased CTRL-Labs, a startup that developed technology that measures neuron activity through a wearable worn on the arm, in order to control digital activity.
Then there’s ongoing medical research, which is more common than you might think.
“This is something that is done today,” Steven Chase, of Carnegie Mellon’s Neuroscience Institute, told Recode. “There are clinical trials ongoing right now where quadriplegic patients have electrodes implanted in their brains, and they use those electrodes and the neural activity recorded on those electrodes to control external devices, such as cursors on computer screens or robotic arms.”
In fact, some of the first medical research into such technology came in the second half of the 20th century, and to some extent, brain-machine interfaces currently exist with limited capabilities. The 1980s saw the invention of both deep brain stimulation and what’s called transcranial magnetic stimulation, which, according to the Mayo Clinic, uses “magnetic fields to stimulate nerve cells in the brain,” and can be used to treat patients with depression.
In the early 2000s came BrainGate, an experimental device that uses an array of electrodes to essentially translate the desire to move limbs from the brain to a device, which is still being researched. The FDA in 2013 approved a system called the RNS Simulator that fires small electrical signals into the brain to stop seizures in some patients with epilepsy.
Already, there are some rudimentary commercial devices that do things one could loosely compare to what a brain-machine interface does. There are headbands that claim to use EEGs to measure your brain activity and then use that data to do anything from enhancing meditation to piloting a drone. These applications are far from the technology Neuralink aims to provide, but they may hint at what our future could look like: Two years ago, DARPA used an experimental brain-computer interface, a surgical microchip, that allowed a paralyzed person to navigate simulated aircraft.
“The idea of sort of sending complex thoughts wirelessly around the world is far, far, far away beyond our lifetime,” said Tim Marler, a senior research engineer at RAND. “This is definitely not science fiction. It will be mature and practical and commercial eventually, but there’s a lot of work to be done still.”
There is a huge technical divide between what’s currently possible in today’s research laboratories and the concept Musk envisions, which requires devices that can handle a significant amount of information going in and out of the brain. One broader hope for brain-machine-interface technology is that it could ultimately help people with paralysis to complete daily tasks on their own. As Chase, from Carnegie Mellon, explained, “The biggest thing these patients want is independence; this technology has the potential to offer them that.”
“As we go deeper into this project, we’re asking ourselves: what is it going to be like to have a thought-based interface? Are you gonna have to think in a different way to send and receive ideas?” remarked Mehin, the designer. “Are you gonna have to train your brain to think a certain way? What’s it gonna be like to receive information?”
But in addition to the technological challenges, the development of brain-machine interfaces also ventures into uncharted ethical and legal territory. On Thursday, the government-funded think tank RAND issued a report on the need for policies around the use of brain-machine interfaces within the military context, where the technology could introduce new concerns like widespread hacking. Of course, with devices that could essentially strip data out of your own mind — including about peoples’ psychological and emotional states — the privacy implications of brain-machine interfaces are also enormous.
“If brain-reading devices have the ability to read the content of thoughts, in the years to come governments will be interested in using this tech for interrogations and investigations,” neuroscience-focused researcher Marcello Ienca told Vox last year.
The list of challenges goes on. Chase raises another concerning scenario: a world in which this technology is only available to the wealthy, creating an extreme technological divide. And then there are the unanticipated health risks of, you know, surgically inserting computer hardware into human brains.
While there’s no reason to worry too much about that right now, Neuralink’s next big announcement is a sign that the idea of connecting human brains to computers on a more regular basis is quickly becoming a reality.
Update Tuesday, September 1, 6:15 ET: This piece has been updated with additional details from Neuralink.