Written by Jeremy Kahn, Jonathan Vanian
Translator: Feng Feng
Source: Fortune Chinese Network
Elon Musk stands next to R1, a surgical robot developed by Neuralink to implant its brain-computer interface device. Musk has convinced Silicon Valley that brain-computer interfaces are a worthwhile business, though many believe he exaggerates the device's capabilities. Image source: COURTESY OF CARD79
At first glance, the video is striking: a monkey is playing the classic Atari game Pong, but only controls it with his mind, and his thinking instructions are wirelessly transmitted to a computer. The video, uploaded to YouTube on April 8, 2021, received 5.7 million views and 120,000 likes.
Elon Musk must be one of the video's iron fans. It's not just because Monkeys Use Their Minds to Play Pong is full of sci-fi gimmicks that excite Musk, but also because the monkey named Pager (from the introduction of the video narrator) is Musk's monkey. Or to put it another way, Pager belongs to Neuralink, a company created by Musk in 2016. Neuralink is working on developing devices that can be implanted into the human brain and translated into action through computers, so that eventually people can do anything they can currently do by tapping a keyboard, clicking a button, or manipulating a mouse or handle, just by thinking about the desired outcome. Musk said it also allows information to be fed back to the brain from the computer. Musk once said that we all need such a brain-computer interface (BCI) in the future, and with the rapid development of artificial intelligence, the only way to keep up with its pace is to integrate with machines, that is, "semi-mechanization".
That sounds a bit outrageous, well, it does. However, the pusher behind it is Elon Musk. He has been more successful than anyone else in promoting the popularity of electric vehicles, and he regularly launches rockets carrying satellites and astronauts into space, with a focus on the rockets that can then be successfully returned vertically to the launch pad for recycling.
Musk became famous because he was able to achieve engineering feats that others thought were technically unattainable and turn them into profitable businesses. Musk also has a bad reputation, though, for overstating his technology and then repeatedly pushing back his promised time nodes [such as Full Self-Driving] and capacity targets [like Tesla's Model 3].
Unlike the layman, many neuroscientists don't find this kind of mind control of Payer as magical, or as excited as Musk tweeted. Since at least 2002, the scientific community has had the ability to implant electrodes in monkeys' brains and teach them how to play games like Pong without their hands. The brain-computer interface already allows monkeys to remotely control the robotic arm and perform some complex actions, and its complexity is far from the controller control action required by Payer to play the game "Pong". You'll be able to find this video online, but you won't see nearly 5.7 million views and 120,000 likes because it's not Elon Musk's monkey.
Neuralink's team isn't the only one working on brain-computer interfaces. For many of these research institutes, they are working on a medical device that can restore some utility and function to patients with spinal injuries and treat Parkinson's disease, multiple sclerosis or schizophrenia. Musk also wants to help such people, but in his opinion, he wants to invent a device that will give those with sound bodies and brains superpowers, so brain-computer interfaces are just the first step towards mass adoption of such devices.
At the same time, there have been signs that Neuralink itself has its own problems. Most of its start-up team members have left the company in the past two years, and employees previously interviewed by Fortune magazine said the company was rife with internal friction and a scruffy management approach. In May 2021, less than a month after The Payer video went live, Max Hodak, a biomedical engineer who has been president of Neuralink since its inception and is responsible for day-to-day management, announced his departure on Twitter. Musk remains the company's titular CEO, but has not succeeded Hodak. The company has so far not received approval from the U.S. Food and Drug Administration for transplanting devices to the human brain, but its competitors achieved that milestone in 2021. (Musk and Neuralink did not respond to numerous requests for comment.) Hodak declined to make any statements. ）
This year, we'll learn more about whether Neuralink can live up to Musk's investment expectations. In December 2021, Musk told the audience: "We hope to apply this device to humans for the first time next year, that is, patients with severe spinal injuries (e.g., quadriplegia, quadriplegia), and the device has entered the FDA approval process." In January, Neuralink released a job posting announcement for clinical trial director, which means the company may be moving toward the timeline that Musk suggested.
For many people in neuroscience and rehabilitation medicine, much of Musk's introduction to Neuralink is problematic because he exaggerates the technology's capabilities. In 2019, Musk said he believed Neuralink's goal was more than just treating brain diseases and cranial nerve disorders. Neuroscientist and technology commissioner James Wu recently tweeted that he was concerned about the excessive attention the company received.
"Neur*link is investing heavily in brain implants that can never be used by ordinary people with disabilities in reality, and this is not a structural shift in our healthcare system," he wrote. (It's speculated that he added the asterisk between Neur and Link to avoid Musk fans pestering the matter by searching for the company's name.) ）
Can Neuralink deliver on Musk's whimsical promises? As Musk's least talked about company, it could become one of the most influential because it helps unlock the secrets of the brain and even changes the nature of "being born to be human."
Neuralink also has the potential to be Musk's most challenging career to date. Christopher Koch, a researcher at the Allen Institute for Brain Science in Seattle, USA, said: "Unlike Tesla and SpaceX, we're not talking about technical issues or infrastructure issues, these are basic science issues. In some ways, the company Musk founded has played an important role: The billionaire's interest in brain-computer interface technology has turned what was previously a niche academic field into a fast-growing commercial industry, similar to Tesla's acceleration of the global electric car space. The question with the Elon Musk effect, however, is: If the technology fails to deliver on musk's boasted promises, will the inevitable shattering of dreams hinder the technology's progress? Will Musk's rhetoric give a false hope to the patients who need a breakthrough in brain technology the most?
However, Utah Array is based on 1990s technology that only covers 100 brain neurons, but the total number of neurons in the brain is about 86 billion, so it will never be possible for humans to merge with artificial intelligence.
In 2016, Musk began quietly forming a dream team to create a brain-computer interface that can greatly promote this technology. He brought together eight "co-founders" of scientists who not only have expertise in brain-computer interfaces, but also have expertise in portable microsensors, brain-inspired computer chips, robotics, and neurosurgery.
Musk has left most of the CEO's key responsibilities to himself. (In the company's registration forms and other documents, Jared Burchar, a wealth manager at Musk's family office, was listed as the company's chief executive; former employees said Burchar had no direct role in Neuralink.) But it was Hodak, the wizard who had worked for the leading brain-computer interface lab created by Miguel Nicolailis of Duke University and was the first to perfect the monkey game of Pong. Although only 28 years old, Hodak has already founded and helped lead several start-ups, including companies that allow researchers to experiment in remote robotic-operated labs.
Nathan Copeland, whose brain-computer interface device was implanted in Blackrock Neurotech, was operating a robotic arm. Blackrock Neurotech's brain-computer interface is based on older technology, but goes further in the medical field than Neuralink. Image credit: COURTESY OF UPMC/PITT HEALTH SCIENCES
For four years, with the exception of unusually rare public recruitment in 2019, most of Neuralink's team has been low-key. The first time the world really saw what Neuralink was building was in August 2020. At a live event, Musk shared the stage with several pigs who had recently undergone neural-computer interface implantation in Neuralink. Musk showed viewers how the data depicted a pig wirelessly transmitting the firing of a pig's neurons to a computer while arching its nose. He also showed a video of a pig walking on a treadmill to demonstrate that neuronal firing data could be used to predict the location of pig joints. The real highlight of the video, though, is Neuralink's hardware, the implanted device called Link in Musk's hand.
The Link is a saucer-shaped device that is comparable in size to the 25 cent coin in the United States, and is small and delicate compared to the Utah Array. Countless ultra-fine wires protruded from it, each thinner than a human hair. When the coin-sized device is placed into a hole in the skull, the threads are buried in the brain. Since the silk thread is capable of moving, it can move with the activity of the brain and hopefully avoid damage to the surrounding tissues. These filaments are also coated with a layer of conductive polymers to cope with the corrosive environment in the brain and minimize inflammation and scarring, which may destroy the ability of brain-computer interfaces to receive signals and may even cause cognitive impairment.
Electrodes are attached to the ends of each branch. Neuralink's senior manager said in an interview with a scientific journal in the summer of 2021 that each Link has 64 branches and 16 electrodes on each branch. This means that each Neuralink brain-computer interface is capable of monitoring 1024 channels, more than 10 times the capacity of the Utah Array. The device processes and filters these signals through its own embedded computer chip, which is a big innovation for brain-computer interfaces that utilize "external" devices for processing. In addition, these links can transmit information wirelessly, and their transmission speed reaches the "megabyte rate" that Neuralink does not explicitly specify (the normal Bluetooth transmission speed is 3 megabytes per second). The device will be completely buried under the skull, "If I had installed the Neuralink device now, you wouldn't have found it, maybe I'd know it myself." Musk said playfully during the speech. The device's battery is capable of being charged wirelessly, which is an important selling point for devices targeting a large-scale consumer market.
To implant a brain-computer interface, Neuralink developed a surgical robot, somewhat similar to a sewing machine, that uses very sharp, very hard tungsten-rhenium alloy needles to locate these tiny threads. The robot drills a hole in the skull and then stitches a filament-like electrode into it — it uses a high-magnification lens to view images of the brain to avoid puncturing a vein or artery — and uses surgical glue to seal the wound. Musk said: "We ultimately hope that this robot can complete the whole operation." He also noted that Neuralink is confident of implanting a Link within an hour without general anesthesia.
With the integration of complex technologies, Link has stunned scientists and a host of laypeople. However, six former Neuralink employees said the process ahead of its 2020 launch was fraught with ups and downs. Most of these former employees have called for anonymity, fearing breaches of confidentiality agreements or strong repression from Musk.
From the beginning, Neuralink's founding scientists were controversial when it came to competing goals. Tim Hansen, a senior scientist at the Janelia Research Campus at the Howard Hughes Medical Institute in Ashbourne, Virginia, said: "Every co-founder has a different opinion. Hansen was a member of the founding team and specialized in the development of brain-computer interface surgical robots for "animal research" at that time. Hansen and other co-founders are more inclined to basic science, while others are more interested in using brain-computer interfaces to treat diseases. Hodak also sees treatment as an important goal, but not its ultimate goal. Musk sees medical use as an aspect of Neuralink, but not its ultimate goal, and he believes that the ultimate goal should be to create a device that can give ordinary people greater ability.
In the words of these former employees, the pressure to deliver results is unprecedented. "Even though we're already moving at an unprecedented pace, the company is not satisfied with the progress of the product from top to bottom." A former tech employee who has worked for Neuralink since 2019 said, "Musk is still not satisfied with this." ”
Multiple employees pointed out that the company's policies are up to Musk, and that these policies prohibit employees from using external suppliers or distributors as an excuse when extensions occur; those who manage the relationship must be held accountable for missing deadlines, even if the reasons for the extension are beyond their control. Former employees noted that this culture of blame and fear led to a high turnover rate in the company.
Given the multiple challenges Neuralink faces, this pressure can pose a number of problems. Hansen said: "There is an imbalance in the company. That is, there is no match between the speed of solving engineering problems and the deliberate acceleration of basic science. Hansen notes: "Progress in basic science is usually slow. He left the company in 2018. Engineers sometimes have to make decisions about issues such as electrode design before the scientific team working on animals can give relevant data. Animal research can take months or years, and engineers are under a lot of pressure because they need to be done in days and weeks. One former employee noted that Neuralink's production of custom-designed computer chips also caused delays. Musk hopes to get into human transplant experiments as soon as possible.
Several former employees said musk, as nominal CEO, whose main focus is on Tesla and Space Exploration Technologies, rarely appears in Neuralink's offices, initially once a week and then twice a quarter, often for hours. Sometimes, he is seen inspecting the hall, usually accompanied by Hodak, former venture capitalist Ben, now Neuralin's director of operations and special projects, Hevern Zillis and a bodyguard. Several former employees said that after Musk's visit, Hodak usually asked them to change their focus, sometimes in a bad manner. One former employee recalled that senior managers had a hard time getting the billionaire's attention when Musk wasn't here, but he had to sign up on major decisions. What made matters even more confusing, the former employee said, was that Musk encouraged a new hire to bypass normal management channels and send him questions and complaints directly via email.
Sometimes Musk intervenes to reject the idea that the team has been working on for weeks, such as telling the team not to use silver metal on Link's electrodes out of concern for potential toxicity, even though the team has planned to wrap the metal silver with a non-toxic polymer. Musk's views on publishing research would also shock scientists: He thinks his engineers should produce products, not papers. However, when other leaders persuaded Musk that it should release a white paper in 2019 that introduced some of its innovations, Musk became its sole signed author, angering researchers.
Musk has a policy in place at his own company against hiring outside contractors and consultants. At Neuralink, the policy spawned decisions that made some former employees feel weird. For example, Hodak initially put Neuralink's head of software in charge of managing the construction of the company's new office in Fremont, California. This huge project required communication with professional contractors and dealing with regulators around complex environmental licensing issues. Hodak said several former employees said the decision was intended to "change the status quo." After months of delays, the company put two managers with mechanical engineering backgrounds in charge of project oversight.
Several former employees said the relationship between Hodak and Musk was growing stiffer, especially after the company's pace was once again behind Musk's target timeline. Finally, on May 1, 2021, Hodak tweeted: "Some personal news: I'm no longer working at Neuralink (left a few weeks ago). I learned a lot there and still have confidence in the company! When someone wrote in a reply: "I'm not in favor of leadership leaving the company before any marketable product has been developed." It seems to have left too early. Hodak replied simply: "Think the same." This response makes it clear that Hodak's departure was not voluntary.
Neuralink has yet to announce a new president. Of Musk's eight co-founders, only two remain in the company: Xu Dongjin and Paul Merola.
Neuralink's former precision machining expert, Felix Deku, recalls that Musk sometimes told employees, "Guys, we need to do this work, people need this instrument." Another former employee said the proactive help often stemmed from emails from desperate patients who thought Neuralink's technology could change their lives, adding to some employees' concerns that Musk was setting expectations too high.
Countless brain-computer experts say Neuralink has brought his discipline into the public eye quickly, and they have received more similar information. Many people are also seeing the rise of other things, such as investor interest. Tech giants like Facebook's parent company Meta, as well as influential venture capitalists like Peter Thiel, have boarded the brains. In a recent report, investment firm Zodiac Capital noted that "Musk has created a fear among investors that should not be missed", and rightfully referred to it as the "Elon Musk effect".
In fact, Musk's interest has helped spawn an entire generation of brain-computer interface startups dedicated to developing products for the medical and commercial sectors. Many praised Musk for taking on an industry that was still far from commercial influence and immediately turned it into an investable area. PitchBook data shows investors poured $531 million into brain-computer interface start-ups in 2021, nearly four times as much as in 2020.
Several members who left Neuralink's founding team are now working for other brain-computer interface companies. Hodak co-founded a company called Science Corporation, which is developing implantable brain-computer interfaces that are less invasive than Neuralink. Benjamin Rappoport, who initially led Neuralink's surgical team, became co-founder of Precision Neuroscience, a startup that calls the products it develops "the least invasive brain-computer interfaces." Philip Sabes, a pioneering brain-computer interface researcher, is now leading a brain-computer interface company called Starfish. Meta hired Vanessa Tolossa to develop hardware for its Reality Labs division, which focuses on navigation devices in the metaverse. Vanessa had developed Neuralink's implant raw materials. "When someone leaves [Neuralink], it's like saying, 'Money for you.'" Deku is not a member of the founding team and joined Brain-Computer Startup Braingrade after leaving Neuralink in March 2021.
Multiple investors told Fortune that this is a long-term investment because they believe it will take years or even decades for the technology to go mainstream. A lot of people are investing in Musk's company. In July 2021, Neuralink announced that it had raised $205 million from well-known venture capital fund firms, including Google Ventures (GV), Vy Capital, DFJ Growth, Craft Ventures, Founders Fund, and Gigafund. Neuralink is also backed by several of Musk's friends, including members of the so-called PayPal Party and Sam Altman, who co-founded openAI, an artificial intelligence company with Musk. Overall, Neuralink has raised $363 million in venture capital funding.
Krishna Yeshvant, a partner at GV, said: "We are at a moment when there may be some huge leap forward in brain-computer interface technology. "He declined to comment on specific companies like Neuralink, but talked about the industry as a whole." I firmly believe that we can help patients to some extent. ”
Even neuroscientists who weren't shocked by Neuralink's presentation have so far been excited about Neuralink's hardware, which they say represents a real leap forward. Koch of the Allen Institute for Brain Science said of Link's device and Neuralink's surgical robots: "It's really remarkable. Reinhold Scheler, co-director of the Brain-Computer Interface Working Group at the University of Essex in England, said Neuralink's hardware innovation and its ability to collect data from numerous neurons over time set an example for the industry as a whole. "Neuralink is doing everything we need to do," he said. ”
At the same time, many people worry that Musk has given people too much hope. In the 2019 demo, Musk said, "If you know someone with a cervical or spinal injury, we're able to solve this problem with a chip." "However, even the most promising outcomes are far from addressing the impact of such harm." Founding team member Hansen said that current algorithms for interpreting brain signals can perform cursor movements and mouse clicks, while more complex commands are still in the laboratory stage. Allowing people with spinal cord injury to regain good motor control is a more difficult challenge: because the brain has no signal feedback from tactile and muscular movements, it is difficult for patients to perform stable movements. Hansen said Neuralink is looking at how to provide this "closed loop" feedback, but its progress will not be fast.
Musk's vision of a full integration of brain-computer interfaces and thinking — a seamless, high-bandwidth bidirectional exchange between the brain and computer — faces even more formidable obstacles. Some are bureaucratic obstacles. Koch said the U.S. Food and Drug Administration is unlikely to approve the transplantation of brain-computer interfaces in healthy humans. He pointed out that it is difficult for any company to prove that cognitive improvements can transcend the risks of transplantation.
Tim Harris, a senior researcher at the Janelia Research Campus and chief designer of brain-computer interfaces at animal research project Neuropixels, writes that the interface Musk seeks requires what neuroscientists call "whole brain coverage," and that to do so, each user needs to implant hundreds of Links. Harris said: "In my opinion, this is a primary issue. How many holes are there for mounting the device and then sensing it? "Even if it can solve the problem, Neuralink will face a bigger problem: how to transmit data back to the human brain? No one knows how to reliably transmit specific information directly to the brain through electrical stimulation. Koch said: "I am 65 years old. In the rest of my years, it was difficult for people to develop techniques close to science fiction. ”
While some people are worried about the impact of the failure, others are wondering, what if Musk succeeds? If you're uncomfortable with Google or Facebook learning your online behavior, imagine how well Neuralink knows about people in the future. Nita Farahani, a law professor at Duke University, said: "The brain is the most sacred area, and it is also the most important place to store privacy and self-creation. If companies and governments, as well as others, could access this information in the future, it would be worrying, if not gruesome. Nita's research interests are the legal and moral dilemmas that come with the development of neuroscience.
Musk has stressed in some public statements that his vision will be difficult to achieve quickly. In a 2019 speech, he said: "It's a slow process. However, he also stressed the urgency of his mission, saying that "even in the slow-moving AI scenario", humanity will become more and more "backward". Then, just as only Musk can do it, he will not hesitate to drill holes in his skull in order to meet the doomsday scene, and call it a seemingly interesting thing. "With the high-bandwidth brain-computer interface, I think we can really go and make a scene."
Smart competition
The "Elon Musk effect" is real: Neuralink's attention has helped spawn a group of startups focused on developing wireless brain-computer interfaces. Here are four of the leaders. ——Jonathan Vanian
Touchpoints. Some transplants rely on tiny needles like those in the Blackrock array (zoomed in). Image source: COURTESY OF BLACKROCK NEUROTECH
Blackrock Neurotech
2022 will be a blockbuster year for the Peter Tille-backed company as it works with the U.S. Food and Drug Administration to pave the way for approval for its latest brain-computer interface device, which itself originates from a device that the U.S. Food and Drug Administration has approved for multiple clinical trials. Neuralink's device can record more brain activity than Blackrock's device, but the former has been in business for more than a decade and has applied its technology to 31 patients worldwide. The technology, based on the 1990s Utah Array device, has helped paralyzed patients regain some of their limb mobility, such as serving a cup of coffee. The next step is to help paralyzed patients complete movements such as moving wheelchairs through thinking.
Synchronous
Synchron's brain-computer interface devices are not implanted in the brain. Doctors would put the device — the Stentrode, an array of electrodes placed on a stent — through a vein in the neck into a blood vessel in the brain that receives electrical signals. The device can help patients who have lost their hands on the ability to move the digital cursor on their laptops. Synchron, who previously implanted the Stentrode device in patients in Australia, has received approval from the U.S. Food and Drug Administration for clinical trials and is looking for its first U.S. subject.
Precision Neuroscience
The co-founders of Precision Neuroscience include Benjamin Rappoport, a founding team member of Neuralink, whose brain-computer interface device is described as a membrane containing thousands of electrodes that can be embedded in small gaps in the skull so that they can be placed at the top of the brain. (The startup's leader describes the process as similar to putting a floppy disk into a disk drive.) Precision hopes to use the device to treat epilepsy and other disorders involving abnormal electrical signals from the brain. The company's founders plan to submit an application for approval to the U.S. Food and Drug Administration in early 2023.
BrainCo
The Harvard University-backed startup has developed a non-invasive brain-computer interface device that looks like a mask from Star Trek. The headband is able to record the discharge activity from the outside through electrodes next to the temple. Although its data isn't as detailed or precise as the data collected by devices inside the brain, the startup executives say the information still helps the device provide feedback on neuronal information so people can train themselves to improve their attention or calm their mood. BrainCo is seeking approval from the U.S. Food and Drug Administration for its BrainRobotics smart bionic hand, which analyzes electrical signals sent to the brain through the arm muscles, giving amputees better control over their movements.
The problem is in your brain
Brain transplant technology has been around for decades, but the biology of the brain itself has been a major obstacle. Here's how Neuralink and other companies are overcoming this difficulty. ——Jeremy Kahn
Brain with transplant = oil and water
For the microelectronic components that help the brain-computer interface work, the environment of the brain is very harsh, and it is composed of a variety of compounds, most of which are corrosive to metals. Destroyed metals release toxic substances. At the same time, the brain does not like foreign bodies to enter it, so it forms scar tissue around the foreign body. This tissue acts as an insulator, reducing the electrode's ability to receive signals. It also has the potential to deflect the electrodes, separating them from the neurons being monitored.
The brain moves frequently, constantly changing positions within the skull and pulsating with blood circulation. If the transplanted electrodes are fixed like most brain-computer interfaces, this movement can damage neurons and blood vessels because brain tissue collides with them in the same way that jellyfish collide with the seawall. The larger the electrodes, the greater the probability of bleeding, especially during the transplant process, a risk that has plagued the larger electrodes that are used for deep brain stimulation in Patients with Parkinson's disease. In addition, there is a risk of chronic inflammation here. In some mouse experiments in the lab, the problem has led to severe cognitive decline.
Bet flexibility
The solution neuralink and other brain-computer interface startups are exploring involves using smaller, more flexible probes to puncture the brain. In Link, each of the 64 silk threads loaded with electrodes is thinner than a human hair. Because these filaments are flexible, they are able to move with the brain instead of constantly colliding with surrounding tissues. Although the thread is made of metal, it is coated with a polymer that strengthens its resistance to corrosion and reduces the risk of damage.
Copyright Notice: In addition to publishing original articles, the manufacturing industry is also committed to the exchange and sharing of excellent articles. Reprint must indicate the source and author of the article; please reprint authorization please leave a message at the end of the article or in the background. All rights reserved, violators will be investigated.