Musk wants to achieve symbiosis between humans and AI through brain-computer interfaces How many brains will be damaged?

Focus:

• 1

  In May, Neuralink announced approval from the FDA to initiate the first human clinical study of brain implants.

• 2

  Because the fear that AI will one day take over the world, helping the paralyzed is not Musk's ultimate goal, and his ambition is to achieve a symbiosis between humans and artificial intelligence.

• 3

  To advance Musk's goals, Neuralink has used potentially dangerous implantation methods.

• 4

  For society as a whole, brain-computer interface human trials also pose ethical risks.

Tencent Technology News October 17 news, foreign media wrote on Monday that among all the achievements of "Silicon Valley Iron Man" Elon Musk - Tesla electric vehicles, SpaceX rockets, Twitter acquisitions, plans to colonize Mars - the mysterious brain-computer interface company Neuralink may be the most dangerous. What exactly does Neuralink do? In the short term, it is to rescue paralyzed patients. But that's not the whole answer. The following is the full text of the article:

Musk funded the founding of Neuralink in 2016. In 2019, the company announced the development of flexible "threads" that will be implanted into the brain by robots and work in a manner similar to a sewing machine. According to Neuralink's vision, flexible "wires" would read signals from the paralyzed patient's brain and transmit the data to an iPhone or computer, allowing the patient to control the computing device with their mind without having to tap, type or swipe the screen.

So far, Neuralink has only tested it on animals. But in May, the company announced approval from the U.S. Food and Drug Administration (FDA) to launch the first human clinical study of brain implants. It is currently recruiting paralyzed volunteers to study whether the implants allow them to control external devices. If the technology works in humans, it could improve the quality of life for millions of people. In the United States alone, there are currently about 5.4 million people who are paralyzed.

But helping paralyzed patients isn't Musk's ultimate goal. This is just one step towards his larger long-term goals. In Musk's own words, his ambition is "to achieve a symbiosis between humans and artificial intelligence." Musk's goal is to develop a technology that helps humans "fuse with AI" so that humans don't "fall behind" as AI becomes more sophisticated. Musk's whimsical idea is certainly unlikely to give the FDA the green light for Neuralink's human trials. But what if it's dedicated to helping paralyzed patients? This will certainly be welcomed by the FDA. And that's just the way it is.

To be clear, Neuralink's technology comes with significant risks. Former Neuralink employees and experts in the field claim that to advance Musk's goal of merging humans with artificial intelligence, the company has enabled an unnecessarily intrusive, potentially dangerous implantation method. The method can damage the brain (apparently this problem has already occurred in animal test subjects). So far, Neuralink has not commented on this report.

For society as a whole, brain-computer interface human trials also pose ethical risks. Currently, many companies are developing technologies implanted in the human brain, capable of decoding consciousness in the human brain and potentially invading mental privacy. We have to be prepared for what's to come.

Why did Musk combine the human brain with artificial intelligence?

Why is Musk's ambition to strive for a symbiosis between humans and artificial intelligence? Because he worries that artificial intelligence could one day take over the world. Today, that fear has spread among AI industry leaders. They generally worry that we may create machines smarter than ourselves, capable of deceiving us and eventually taking control from us.

In March, many bigwigs, including Musk, signed an open letter calling on all AI labs to immediately suspend the training of AI systems more powerful than GPT-4 for at least 6 months. The letter warned: "Should we develop non-human brains so that they eventually surpass human numbers, outperform human intelligence, and eliminate and replace humans?" Should we risk losing control of our civilization? Such decisions must not be delegated to unelected technical leaders. It should only be developed if we are confident that the effects of powerful AI systems are positive and that their risks are manageable. ”

While Musk isn't the only big guy warning about the "civilization risk" posed by AI systems, what sets him apart from the rest is that he's quietly devised plans to mitigate them. The basic logic of Musk's plan is: if you can't defeat your opponents, join them.

In the future world Musk foresees, artificial intelligence systems will be able to communicate information at 1 trillion bits per second, completely despising humans who can only communicate at 39 bits per second. Humans seem useless to AI systems at this point. Unless, humans become like them. In Musk's view, a big part of that is being able to think and communicate at the speed of artificial intelligence. "This will involve the bandwidth and connection speed between the human brain and the digital version of itself, especially in terms of output," Musk said in 2017. "The high-bandwidth interfaces that will be able to connect to the human brain in the future will be the things that can help humans achieve symbiosis with machine intelligence, and can solve the problem of control and effectiveness."

Nearly 6 years after Musk made this speech, we can see that he is still obsessed with the concept of bandwidth, which is the rate at which computers read information from the human brain. In fact, it also drove Neuralink's vision. Neuralink's device is a brain implant equipped with 1,024 electrodes that can pick up signals from a large number of neurons. The more electrodes, the more neurons that can listen, and the more data you get. In addition, the closer to the neuron, the higher the data quality.

Neuralink's device is very close to neurons. The procedure the company implanted it entailed drilling a hole in the skull and penetrating the brain. There are some less extreme ways to solve this problem, and other companies are proving it. Let's analyze what these companies are doing and why Musk feels the need to do something unusual.

Why did Neuralink choose the most extreme brain-computer interface access method?

Neuralink isn't the only company exploring the use of brain-computer interfaces to restore motor function in paralyzed patients. Companies such as Synchron, Blackrock Neurotech, Paradromics and Precision Neuroscience, as well as the U.S. military, are exploring this area.

In recent years, many of the headlines have focused on brain transplants that translate the minds of paralyzed patients into language. Mark Zuckerberg's Meta, for example, is working on brain-computer interfaces that can extract ideas directly from human neurons and translate them into language in real time. In the long term, the company says, the goal is to enable everyone to use their own ideas to control hardware devices such as keyboards and augmented reality headsets.

Early successes in the field of brain-computer interfaces focused on motion, not language. In 2006, a quadriplegic patient named Matthew Nagle in Massachusetts, USA, successfully controlled televisions, computers, robotic arms and other devices with his mind after undergoing brain-computer interface implantation surgery. Shortly after surgery, Nagel was able to play the first arcade game in history, Pong, with his mind.

The device implanted in Nagel's brain, developed by research institute BrainGate, uses the Utah Array, which surgically implants a series of 100 pointed electrodes into the brain. Although the Utah electrode array is about one-tenth the size of the electrodes in a Neuralink device, it still allows paralyzed patients to move their cursors with their minds, check emails, adjust TV volume or channels, and control robotic limbs. Since then, other paralyzed patients have also recovered somewhat through brain-computer interface technology.

Early technologies like the Utah array would stick awkwardly out of the skull, but once the latest brain-computer interface is implanted, it is not even noticeable to outside observers, and it is much less invasive. For example, Synchron's brain-computer interface technology is built on scaffold technology that has existed since the 80s. Metal stents can be introduced into blood vessels, allowing them to remain safely in blood vessels for decades (many heart patients have long adopted this technique to keep their arteries open). Synchron uses a catheter to deliver a stent into a blood vessel in the brain's motor cortex. Once in place, the scaffold unfolds like a flower, and the sensors on it receive signals from neurons. The technology enables some paralyzed patients to tweet and text messages with their minds.

In other words, existing brain-computer interface technologies require neither craniotomy nor drilling holes in the skull. At the 2016 Code conference at Recode, Musk himself said in a 5-minute video that brain-computer interfaces do not necessarily require craniotomy. "You can go through veins and arteries because that provides a complete channel for all your neurons," he said. "You can insert something in the jugular vein and..." After the audience laughed nervously, he added: "It doesn't involve cutting off the head or anything like that." ”

Several former Neuralink employees revealed that one of the company's research and development teams considered gentler vascular implantation methods when it was founded. The team explored ways to deliver devices to the brain through arteries and demonstrated that this is possible. But by 2019, Neuralink abandoned that option in favor of a more invasive surgical robot that implants flexible "threads" directly into the brain.

If vascular implantation can restore critical function to paralyzed patients while avoiding some of the safety risks that come with crossing the blood-brain barrier, such as inflammation and scar tissue buildup in the brain, why would Neuralink choose a more invasive method than this method?

While Neuralink didn't explain this, according to Hirobumi Watanabe, who led Neuralink's vascular implantation research team in 2018, the main reason is the company's obsession with maximizing bandwidth. "Neuralink's goal is to implant more electrodes to be able to collect more data from more neurons," Hirobumi Watanabe said.

After all, Musk has proposed that seamless integration with machines could allow us to do anything from enhancing memories and uploading ideas to immortality — all staples of Silicon Valley's transhumanist fantasies. This may help to understand Neuralink's dual mission: "to create a universal brain interface to help people who currently do not have access to medical care, and to unlock human potential in the future." ”

"Neuralink's clear goal is to produce universal neural interfaces," says Marcello Ienca, a neuroethicist in Munich. "As far as I know, they are the only company currently planning clinical trials for implantable medical neural interfaces. At the same time, the company also issued a public statement saying that cognition-enhancing neuroimplants will be used in non-medical fields in the future. To create universal technology, a seamless interface between humans and computers needs to be developed to enable enhanced cognitive and sensory abilities. Achieving this vision may indeed require a more intrusive approach to higher bandwidth and precision.

Hirobumi Watanabe believes that Neuralink prioritizes maximizing bandwidth because it aligns with Musk's goal of creating a universal brain-computer interface that allows humans and artificial intelligence to converge. "Because that's what Musk said, that's what Neuralink has to do," he said.

Endovascular implantation doesn't seem to provide as much bandwidth as a cranial drill. While it may be safer to stay in the blood vessels, the downside is that it doesn't have access to as many neurons. "That's the biggest reason Neuralink hasn't taken this approach," Hirobumi Watanabe said. "It's sad." He added that he thought it was a bit too early for Neuralink to abandon the minimally invasive approach. "We could have moved forward with this project."

For Tom Oxley, Synchron's CEO, this raises a big question. "The question is, will there be a conflict between the short-term goals of patient clinical health outcomes and the long-term goals of AI symbiosis?" He said. "I think the answer is probably yes. What matters is what the original intent of the design is and whether the patient's problems are considered," Oxley added. Synchron could theoretically increase bandwidth by miniaturizing its technology and penetrating deeper branches of blood vessels; And studies have shown that it works. "But Neuralink chose the other extreme approach," he said.

Ben Rapoport, a neurosurgeon who left Neuralink to start Precision Neuroscience, emphasized that any time electrodes penetrate the brain, it is causing some damage to brain tissue. If the goal is to help a paralyzed patient, this approach is completely unnecessary. "To restore speech and motor function in patients with stroke and spinal cord injury, I don't think this neuroprosthetic function is necessary. One of our guiding ideas is that high-fidelity brain-computer interface systems can be built without damaging the brain. ”

To demonstrate that Musk's invasive implants aren't needed to achieve high bandwidth, Precision designed a thin film that covers the surface of the brain with 1,024 electrodes — the same number as in Neuralink's implant — that can transmit signals similar to Neuralink's implants. This film must be inserted through a crack in the skull, but the advantage is that it is located on the surface of the brain without penetrating the brain. Rapoport calls it the "Golden Phoenix Girl (just right) solution." Precision has implanted the film into the brains of a small number of patients, recording their brain activity in high resolution.

"The key is to have a very, very safe procedure that doesn't damage the brain and is minimally invasive in nature," Rapoport said. In addition, as we expand the bandwidth of the system, the risk to patients should not increase. It makes sense if your most cherished ambition is to help patients improve their lives as much as possible without taking unnecessary risks. But we know that Musk has other ambitions. ”

"While a more intrusive approach may provide an advantage in terms of bandwidth, it raises greater ethical and security concerns." Neuralink doesn't seem to care about that," Rapoport said. "At the very least, I haven't heard any public statements about how this company intends to address the greater privacy, security and mental integrity risks arising from their approach." This is strange because, according to international research ethics guidelines, it is unethical to use less invasive techniques to achieve the same performance if less invasive methods are used. "As Neuralink's experiments in animals have shown, more invasive methods, by their very nature, can cause real damage to the brain.

Ethical questions raised by Neuralink

Some Neuralink employees revealed that because the company rushed to perform botched surgeries, the pigs and monkeys used in the experiment suffered more than necessary and died. They claim that after Musk repeatedly claimed that he would begin human trials as soon as possible, he had been pushing employees to get approval from the U.S. Food and Drug Administration as soon as possible.

For example, Neuralink implanted 25 of 60 trial pigs with missized devices in 2021. Later, the company killed all the affected pigs. Neuralink employees told the media that the mistake could have been avoided if they had had more time to prepare.

Veterinary reports indicate that the monkeys Neuralink used for the trial also suffered a terrible fate. In one monkey, part of the device "fell off" during implantation in the brain. The monkey scratched and dragged until part of the device came off, and then the infection began. Another monkey had bleeding in her brain, and the implant left parts of her cortex "tattered." Eventually, both monkeys were euthanized.

In December, the USDA Inspector General's Office opened an investigation into Neuralink's treatment of some animal subjects. The company is also facing an investigation by the Department of Transportation, which concerns that implants taken from the brains of test monkeys could be unsafely packaged and transported, potentially exposing people to pathogens.

"Neuralink's past animal testing has demonstrated serious safety concerns stemming from the invasiveness of the product and the hasty and hasty actions of company employees," the Physicians Committee for Responsible Medicine, a nonprofit that opposes animal testing, said in a statement released in May. "In view of this, the public should continue to be skeptical about the safety and functionality of all equipment produced by Neuralink."

What if Neuralink's approach works too well?

In addition to what surgery means for individuals recruited to participate in the Neuralink trial, there are ethical questions about what brain-computer interface technology means for wider society. If the high-bandwidth implants Musk is pursuing really allow unprecedented access to what's going on in people's brains, that could increase the likelihood of dystopia. Some neuroethicists argue that the potential for abuse is so great that we need to amend human rights law to protect ourselves before moving forward.

First, the brain is the last privacy boundary. The brain is home to human personal identity and the most intimate thoughts. What happens if the precious 3 pounds of mucus in the skull is not controlled by the humans themselves? Imagine a scenario where the government uses brain-computer interfaces for surveillance or interrogation. If authorities have the right to eavesdrop on individuals without their consent, the right not to incriminate themselves under the U.S. Constitution may become meaningless.

Second, experts are also concerned that the device Neuralink is making could be vulnerable to hacking. What happens if a user is using a brain computer and a malicious actor intercepts the Bluetooth connection, altering the signal entering the brain, making the user frustrated or submissive? Neuroethicists call this brain hijacking. This is still hypothetical, but this possibility has already been demonstrated in proof-of-concept studies. Hacking like this doesn't require that sophisticated technology.

Finally, consider how the user's mental continuity, or basic sense of self, can be imposed or disrupted by brain-computer interfaces. In one study, an epilepsy patient with a brain-computer interface began to feel a complete symbiosis with it. The patient said: "It became me. Later, the company that implanted the device in her brain went bankrupt and she was forced to remove the brain-computer interface. She cried, "I lost myself. ”

To fend off the risks of a hypothetical almighty AI in the future, Musk wants to create a symbiotic relationship between the human brain and machines. But this symbiotic relationship also generates its own very real risks, and these risks are now befalling humanity. (Mowgli)