Can Neuralink really put AI in our brains?
As Neuralink announces human trials, are we ready for AI-augmented humans?
Elon Musk likes to be perceived as a pioneer on the frontier, a rogue entrepreneur straddling an AI-powered robotic horse wielding a satellite-guided laser pistol in each hand as he gallops head-first into the unknown. Musk’s Wild West is the colonization of Mars, artificial intelligence, solar energy solutions, reusable spacecraft, a constellation of internet-providing micro-satellites, gigantic transportation tunnels under major cities, electric vehicles, and the brain-computer interface (BCI) universe. In many domains, his companies are pushing boundaries. At the end of January 2024, the BCI start-up he co-founded, Neuralink, officially announced the first implant of a computer chip in a human test subject.
Neuralink’s implant of a brain-computer interface device in a human follows the previous breakthrough the company announced in 2021: the successful implant of the ‘N1 chip’ into the brain of the pong-playing, 9-year-old macaque, Pager.
Pager’s ability to control the game with his mind is certainly impressive, but this mildly heartwarming yet mostly terrifying video leaves out the context of how Neuralink achieved these results. It wasn’t a painless accomplishment, nor was it transparent or ethical.
‘On’ or ‘in’ the brain
Neuralink is not the first company to implant chips into human brains, nor has it assembled the only group of scientists working on the challenge. “A start-up called Synchron backed by funding from investment firms controlled by Bill Gates and Jeff Bezos, has already implanted its stent-like device into 10 patients.”
Other competitors include Kernel, Paradromics, Nuro, Blackrock Neurotech, and BrainUp, among others. They are all working on a version of Brain-Computer Interfacing, some with less invasive methods than others, with various stated goals and applications.
Judging by the objectives and achievements of Neuralink and its competitors, there are two options for computer interfacing with the brain inside the skull (non-invasive solutions exist, consisting of reading neural activity outside of the skull with a helmet-like device). The receptors of the implanted device are either placed “on” or “around” the brain to collect an aggregate reading of neural activity (Synchron’s method), or they are placed “in” the brain to pick up readings from individual neurons (Neuralink’s method). This means that for Neuralink’s process to work, receptors must be inserted as close as possible to individual neurons inside brain tissue.
The N1 Implant records neural activity using 1,024 electrodes. Each of these 1,024 electrodes is a “highly flexible, ultra-thin” filament that’s thinner than a human hair. That means 1,024 sensor filaments must be inserted into the brain tissue of each Neuralinked brain. Neuralink has even designed a robot to conduct the surgery that would be too complicated for a human hand to perform.
In comparison, the Synchron device is “inserted through the patient’s blood vessels” and “delivered to the large vein that sits next to the motor cortex.” “Since Synchron’s BCI isn’t inserted directly into the brain tissue, the quality of the brain signals is not as strong, according to the company.”
Neuralink and its competitors are pushing the message of noble intentions in the discourse about their products: giving people with neurological conditions and paralyzed people the ability to control a computer with their mind, and even to maybe move limbs or walk again. The initial stated goal for Neuralink is to create technology that will give “people with quadriplegia the ability to control their computers and mobile devices with their thoughts.” But beyond assisting the lives of patients with neurological diseases and paralysis, the company’s mission statement is more ambitious and broader in scope, which is to “create a generalized brain interface to restore autonomy to those with unmet medical needs today and unlock human potential tomorrow.”
Dead animals & opaque science
A damning report by WIRED contradicts the statements Musk made about the health and safety protocols in place during the monkey experiments Neuralink carried out in its initial testing phases. Before the company could seek authorization for human trials, it had to prove the concept worked in animals.
All of the monkeys involved in the early tests had to be euthanized and Musk claimed that the “monkeys who died during trials at the company were terminally ill and did not die as a result of Neuralink implants.” He “denied that any of the deaths were ‘a result of a Neuralink implant’ and said the researchers had taken care to select subjects who were already ‘close to death.’”
But veterinary reports that have since been made public “contain gruesome portrayals of suffering reportedly endured by as many as a dozen of Neuralink’s primate subjects, all of whom needed to be euthanized.”
“Public records from the University of California, Davis, where Neuralink conducted monkey experiments between 2017 and 2020, reveal that implantation of the company’s device caused debilitating health effects in monkeys, resulting in euthanasia. Animals experienced chronic infections, paralysis, swelling in the brain, loss of coordination and balance, and depression.”
The article details the gruesome conditions the monkeys faced in their final days. A former Neuralink employee told WIRED that “up to a year’s worth of behavioral training was necessary for the program, a time frame that would exempt subjects already close to death.” When shown a copy of Musk’s remarks on X about Neuralink’s animal subjects being “close to death already,” the former Neuralink employee told WIRED the claim is “ridiculous,” if not a “straight fabrication.”
Reuters reported in late 2022 that Neuralink had “killed about 1,500 animals, including more than 280 sheep, pigs and monkeys, following experiments since 2018,” according to sources with “direct knowledge of the company’s animal-testing operations.” The figure is a rough estimate since Neuralink “does not keep precise records on the number of animals tested and killed.” In contrast, Synchron also conducted tests on animals, but it killed only about 80 sheep as part of its research, according to studies of the Synchron implant reviewed by Reuters.
The excessive animal deaths seem to be a direct result of Musk’s desire to make progress as quickly as possible in getting Neuralink approved for human trials. Neuralink staff blew the whistle on their working conditions and in late 2022, Neuralink was “under federal investigation for potential animal-welfare violations amid internal staff complaints that its animal testing is being rushed, causing needless suffering and deaths.” According to Neuralink employees, “pressure from CEO Musk to accelerate development has resulted in botched experiments” and those “failed tests […] had to be repeated, increasing the number of animals being tested and killed, the employees say.”
Through company discussions and documents spanning several years, along with employee interviews, Reuters identified four experiments involving 86 pigs and two monkeys that were marred in recent years by human errors. The mistakes weakened the experiments’ research value and required the tests to be repeated, leading to more animals being killed, three of the current and former staffers said. The three people attributed the mistakes to a lack of preparation by a testing staff working in a pressure-cooker environment. One employee, in a message seen by Reuters, wrote an angry missive earlier this year to colleagues about the need to overhaul how the company organizes animal surgeries to prevent “hack jobs.” The rushed schedule, the employee wrote, resulted in under-prepared and over-stressed staffers scrambling to meet deadlines and making last-minute changes before surgeries, raising risks to the animals.
The suffering monkeys are the tip of the iceberg. The ethical quandaries Musk and his team are wrestling with are numerous and complicated. Not least because the data from Neuralink’s human experiments is intentionally withheld from the public. “Neuralink has previously refused National Institute of Health (NIH) funding to keep its patented technology secure as federally-funded research comes with more oversight and mandated requirements of data sharing and transparency. While most academia-sponsored clinical trials that happen through major university hospitals in the US need to be registered (at clinicaltrials.gov), the PRIME study (as christened by the company) is not. The only information available is through an online brochure, which is characteristically minimalistic.” “Because the trial is unregistered, it is virtually impossible to find out anything about the conditions under which it will be conducted, commonly referred to as a study protocol.”
Musk is navigating a delicate balance, juggling scientific advancement with investor relations. He must demonstrate the safety of Neuralink's brain tissue implant for large-scale production, ensure that its capability to interpret neural activity remains stable over time, and establish a robust plan to safeguard the data generated by each Neuralink-connected brain. He can’t afford to lose the confidence of investors to whom he has promised “both telepathy between people and advantageous relations with artificial intelligence.” The start-up has raised “a total of $323 million from 32 investors” and was valued at $5 billion in June 2023.
The Physicians Committee for Responsible Medicine (PCRM) said in September 2023 that, “Because the invasiveness of Neuralink’s device poses serious health risks to patients, Musk is misleading investors about the safety and marketability of the company’s device.”
Beyond the concerns about publicly unavailable data, “the threading of up to 1,024 recording electrodes in the brain will lead to some form of micro-injuries in the brain and Neuralink will need to convince the scientific community that these implants do not enhance the likelihood of microbleeds or strokes or any other form of injuries in the brain. Such microscopic injuries are often not picked up by any diagnostic modalities yet they accumulate over time and predispose individuals to serious neurological conditions such as neurodegeneration later on. The study has to prove that the recording capacity of the electrodes is maintained consistently—this is a problem with existing BCIs. The efficiency of the recording decays over time. The PRIME study calls for 18 months of primary observation followed by periodic follow-ups up to five years.”
The absence of registration in a clinical trial repository and the lack of transparency regarding the study protocol also cast doubt on the legitimacy of publishing these results. It's worth noting that numerous reputable medical journals now exclusively publish findings from registered trials.
One key ethical consideration revolves around data ownership. The question of who rightfully possesses the recorded data, which is used to discern an individual's 'intentions' for controlling a digital device, remains unclear. The significance of this issue might be underestimated for now, given that the trial's subjects are clinically disabled and stand to directly benefit from a BCI implant.
However, Neuralink's aspirations transcend clinical applications to augment human cognition and potential. Considering this broader scope, the ethical implications of data ownership are significant and require transparent solutions. Establishing clear guidelines from the outset is paramount, as any missteps could pose substantial risks to human autonomy and privacy on an unprecedented scale.
Human-AI symbiosis
The smorgasbord of ethical issues such an invasive product entails seems to be a trivial concern for Musk, who “argues that the Neuralink device would empower humans to compete with emerging sentient AI, stating, ‘I established Neuralink specifically to address the AI symbiosis problem, which I believe poses an existential threat.’”
The key takeaway from Musk’s push into the realm of the semi-human, semi-machine world is his belief that humans must become well-equipped cyborgs if we are to survive the advent of super-intelligent, super-capable artificial intelligence. He thinks AI is “one of the existential risks that we face, potentially the most pressing one.” The solution for Musk is achieving symbiosis between “our biological mind and our digital mind,” in essence merging our brains with computer intelligence, overcoming some of the latency that exists in the input/output gap in the communication between biological beings and machines.
Musk talked about eliminating the “input/output constraint” in 2016, the year Neuralink was launched, suggesting that the biggest hurdle to achieving human/digital intelligence symbiosis is overcoming slow communication. Using our fingers to output from our brain into the computer is a “ridiculously slow” method of transferring data. Enter Neuralink to jump over this I/O hurdle, enabling real-time human/digital intelligence communication and ushering in AI-augmented humans.
Musk’s rush to develop Neuralink is an indicator of the paradigm shift he believes our species is about to face, transparent science and animal welfare be damned. In early 2023, Musk and other tech leaders signed an open letter “calling for artificial intelligence labs to stop the training of the most powerful AI systems for at least six months, citing ‘profound risks to society and humanity.’”
Humanity is caught in yet another paradox. “Top executives at some of the biggest tech companies are telling the public that AI has the potential to bring about human extinction while also racing to invest in and deploy this technology into products that reach billions of people.”
Will artificial intelligence outpace humanity’s ability to keep it constrained and benevolent? Are we following Musk into the unknown of the Wild West despite the moral quandaries of the journey for a chance to save the species from AI-wrought extinction? Are we simply spectators to an inevitable fate, having nearly completed our role as the sex organs of the machine world that was always destined to replace biological life? Or have we deviated from solving the very real human problems that confront our species daily, distracted by gadgets that tickle our sci-fi fantasies?
Time will tell.
