Will Elon Musk’s Neuralink be able to deliver on its many promises?
The eccentric billionaire has made some grandiose statements regarding the possible uses of his brain-computer-interface product, yet, the scientific community remains unimpressed
Now, every time Gertrude eats or sniffs hay, the chip sends wireless signals to a computer, which beeps and registers the pig’s brain activity on a graph.
This may sound like a mundane occurrence, but if you believe the promises made by the eccentric entrepreneur, it is nothing short of a revolution that will pave the way to great things: from accurately controlling robotic limbs through restoring vision to the blind and hearing to the deaf and up to notions that sound more like science fiction, such as reading minds, automatic universal translation, and the ability to accurately record and retrieve memories.
Musk founded Neuralink in 2016 and since then the company has operated under a veil of extreme secrecy. All that is known about its technology is that it is based on a chip that has far more electrodes than the industry standard, which is implanted in the brain to allow a more accurate reading of its activity.
The company also developed a brain surgeon robot that operates similarly to a sewing machine to implant the chip.
Musk is known for his belief that artificial intelligence poses a threat to human society. Now, Neuralink’s mission statement, Musk declared in a tweet, is “if you can’t beat em, join em.”
If you can’t beat em, join em— Elon Musk (@elonmusk) July 9, 2020
Neuralink mission statement
In his press conference, Musk referred to Neuralink’s future abilities, stating it could allow humans to save and replay memories. The similarity to a popular cultural phenomenon did not escape him as he stated “this is obviously sounding increasingly like a ‘Black Mirror’ episode.” They must be “pretty good at predicting,” he concluded.
The notion of a brain-computer-interface (BCI)—a direct communication line between the brain and an external device, is nothing new. It can be achieved through invasive means, such as Neuralink’s implants, or non-invasive and, as a result, less accurate means, such as the wristband developed by Facebook subsidiary CTRL-Labs (incorporated as Cognescent Corp.), which measures electrical pulses using electrodes placed on the skin to decipher instructions from the brain.
The birth of BCI as a research field is attributed to the establishment of the BrainGate team in the early 2000s, in which neurologists, brain scientists and surgeons, computer scientists, and mathematicians work together to develop a technology that would preserve and restore the communication and movement ability of people who have a neurological disease, injury, or amputated limbs.
The team is affiliated with some of the top medical and research institutions in the U.S., including Brown University, Massachusetts General Hospital (MGH), Providence VA Medical Center, and Stanford University. In 2004, the team received a green light from the FDA to conduct clinical trials for its brain implanted sensor. The sensor detects brain signals and deciphers them, letting users control a computer cursor.
The first trial was done on four patients, one of whom was Matt Nagle, a 25-year-old former athlete who became paralyzed from the neck down after being stabbed in the back. A 96-electrode chip called the "Utah Array" was implemented in Nagle’s brain in June that year. With the implant, Nagle was able to move a computer cursor with his mind and send basic commands to an external prosthetic limb.
“I can't put it into words,” Nagle, who later died of sepsis caused by his injury, was quoted as saying at the time. “I just thought it. I said ‘cursor go up to the top right.’ And it did and now I can control it all over the screen. It will give me a sense of independence,” he added.
The encouraging results of the first tests paved the way to a second clinical trial that was launched in 2009 and includes nine patients. The research is still ongoing but some of the results were published in 2012 by British scientific journal Nature.
According to BrainGate’s findings, two of the patients managed to control robotic arms with their minds and grab objects and one of them even managed to have her first latte without assistance in 15 years.
BrainGate continues to operate as a research team without a commercial arm but its workforce has changed dramatically since the establishment of Neuralink. This should come as no surprise as Musk, whose personal fortune is estimated at over $93 billion, went on a shopping spree resulting in the recruitment of many former BrainGate employees who are now part of Neuralink’s team of 100 people. The explicit purpose of Musk’s recent press conference was to recruit more employees, so, it is likely this already talent-crunched industry will experience more transition waves in the near future.
Neuralink’s main competitor is Austin-based Paradromics Inc. Founded in 2015, Paradromics has raised just $25 million to date, small change compared to the $158 million invested in Neuralink, $100 million of which came directly from Musk’s pockets.
Paradromics’ flag product, Argo, is an 8 square millimeter chip, with 30,000 electrodes, that is meant to be implemented in the cortex for medical uses. The chip is currently at the prototype stage and development is expected to take another year.
As soon as it will be possible to transfer information directly to the brain, many common medical conditions could be referred to as a data transference problem, Paradromics founder and CEO Matt Angle told Calcalist in an interview. Blindness is nothing but the inability to transfer visual data to the visual cortex and the visual system, Angle said, and deafness is the inability to translate sound into neural activity in the auditory cortex.
Similarly, Angle suggested, it is possible to view paralysis from a data transfer perspective. “It is as if a line has been cut,” he said, “the brain creates the signals but the muscles cannot receive them.”
Unlike Neuralink’s product, Paradromics’ chip is implemented by a human surgeon, who drills a tiny hole in the skull, removes the protective tissue of the outer layer of the brain, and places the implant. According to Angle, the procedure is much like a craniotomy, an operation typically performed for various neurological conditions.
Unlike Musk, who has compared Neuralink’s procedure to laser eye surgery in terms of the level of risk and complexity, Angle is more cautious. “The risk is always infection,” he said, “it exists every time you open the skull and expose the brain. But, when it comes to a blind, deaf, or paralyzed patient, or one who suffers from serious mental illness, the benefit outweighs the risks.”
Another prominent player in the field is Synchron Inc. which has just 10 employees and develops a completely different type of BCI technology. Its product, called Stentrode, is a small device that is injected into the bloodstream and intercepts the brain’s signals, as it moves through it.
The procedure resembles the installation of a pacemaker and takes up to two hours under local anesthesia. The upside is clear—there is no need to open the skull. The downsides, however, are also evident. First of all, as the device is not physically connected to the brain, it might prove to be less accurate.
Also, it is yet unclear what the long term effects of devices moving through blood vessels for long periods of time would be. That is why, for now, patients are treated with Aspirin, for blood thinning, in order to reduce the possible risk of blood clots.
Synchron is considered the most advanced of the three in terms of human trials and regulatory approvals. A day before Neuralink’s press conference, Synchron announced it has received an FDA breakthrough device designation. In the upcoming weeks, it is expected to publish the results of its first clinical trial and start raising a second funding round. To date, it has raised $25 million.
The company intends to conduct a second clinical trial on 15 patients, hoping to gain the approval to bring its product to the market, founder and CEO Thomas Oxley told Calcalist. Assuming these trials are successful, Stentrode will hit the shelves in about five years.
What if Neuralink beats you to it?
“I don’t see it as a competition. We are all just trying to solve an existing problem and I find it exciting that Musk is putting his energy into it. I think this will help the industry develop and that is very exciting for patients with paralysis that are just waiting for this type of technology to turn commercial. I don’t know enough about Neuralink's science to comment on it, but Musk can build rockets and electric cars and I am certain he will build a very convincing product.”
Angle also believes Musk can help the industry, which is currently underfunded. “Sometimes, when you are working on something that truly excites you and is at the forefront of research but you can’t draw attention to it because it is very esoteric,” he said, “it can be very frustrating when someone else gets such a crazy amount of media attention. Especially if you feel that what you are doing is essentially far more advanced.”
On the other hand, Angle said, just consider how much the world has heard about BCI since Musk entered the field. His involvement, Angle explained, has raised the public’s interest in the work of other people as well, which has created more private investments as well as funding opportunities from the government.
“What Neuralink presented,” Angle said, “may not have been a scientific innovation but it did combine some great practices in the device’s development and it is on its way to becoming an actual product. This type of integration work may not win you scientific praises but it is much needed and very important.”
The passive-aggressive remarks by Angle and Oxley are typical of the sentiment about Neuralink and Musk within the scientific community. An adult human brain has more than 80 billion neurons, each connected to around 10,000 similar cells. This means that despite Musk’s grandiose statements, the mapping of these connections is just in its early days and it will be very long before we are able to read minds or record memories.
Neuralink appears to have made a significant step in designing the hardware for intercepting and registering signals in the cortex at a relatively high level of accuracy compared to the industry standard, biomedical engineer Bolu Ajiboye of BrainGate told Calcalist.
But, Ajiboye added, there are still quite a few unanswered questions, concerning, for example, the product’s durability, how stable its signal registration is, and what the long term biological response of the brain to the implant would be. “There are still too many things we do not know in terms of how the brain works, both in Neuralink’s research and in neuroscience in general,” he said.
Rony Paz, the head of the neurology department at Israel's Weizmann Institute of Science, feels the same way. “Nobody knows what is going on at the heart of Neuralink’ research, which is yet to be disclosed, but at least in what Musk presented in the event, there is no innovation in terms of our understanding of the brain, the way it works, and how psychiatric or neurological disease can be treated,” he told Calcalist. “He did not present a breakthrough or any progress in one of these aspects, instead, just an impressive technological development—a chip that has more electrodes and is slightly less invasive.”
Neuralink is like a company that develops a very powerful telescope lens that can be attached to a satellite, Paz said. “This could teach us a lot about astrophysics,” he said, “but the knowledge is not going to come from the company that only knows how to build lenses, it will come from astrophysicists who will use the data collected through the lens to gain a better understanding of the laws of physics.”
Also, Neuralink’s chip attaches to the cortex, which means it only has access to the most outer layer of the brain, so it is very limited, Paz said. There are already experiments being made on animals to implant electrodes in other parts of the brain, he said, adding he believes Neuralink will get there too.
So, what is Neuralink’s product good for?
"The fact that it has more than 1,000 electrodes per chip, instead of several dozens or hundreds, will help it be more accurate, for example, when it comes to moving robotic arms.
“Also, the engineering method—saying ‘I don’t need to understand how it works, I am just looking for a solution to this problem’—sometimes proves more effective than a scientific approach to a specific problem. In other words, you can make trial and error experiments even without a deep understanding of how the brain works.
"It is possible that Musk and his researchers will find something that works and could help patients suffering from depression, for example. Naturally, we all want that day to come but there is a huge difference between that and reading minds or implanting thoughts. In order to create a new memory or read an existing one, you have to really understand how the brain works and we are lightyears away from that.”
So we won’t be able to summon our Tesla with our minds any time soon?
“Actually, that would be relatively easy with the engineering method. All Musk would have to do is ask anyone coming in to buy a Tesla to think about Tesla 50 times. Once he did that he would have a controlled experiment and could know which electrode activity correlates with the person thinking about Tesla. This could be encoded onto the chip and anytime this particular activity would pop up on its radar, the Tesla would emerge.”
Alongside the promise of an improvement in the quality of life for people with paralysis or disabilities, BCI also raises some ethical issues and privacy concerns. The ability to decipher thoughts could prove to be a slippery slope.
“In the short term,” Paz said, “the question is why and how implanting electrodes in human brains will be approved. Will they only allow it for the treatment of clinical conditions or will it be further implemented due to political pressure or public opinion?
“You cannot develop technologies that have the potential to make things better without them also being dangerous when not used right. There needs to be a delicate balance here, as the ethical concerns are grave, much like issues concerning genetic data.”
On top of these issues, Neuralink’s name has also been linked to some questionable practices. Several days before the press conference, health news website Stat published a reportage on Neuralink outlining a chaotic internal culture placing speed over professionalism.
According to the report, the company was haste to test a new and dangerous operation on monkeys and considered bypassing the long and rigorous regulatory process in the U.S. and beginning human trials in China or Russia. The company denied the claims.
Musk’s extravagant and daring management style has helped him build several empires but the medical research field requires one to be over-cautious, which means seeing slower progress.
You can get cars to drive themselves and rockets to fly out to space relatively quickly, but when traveling into the depths of the human brain, one needs to take a deep breath and remember there is still a very long way to go.