The Starset SocietyConnecting the human brain to an artificial system
By: A’liya Spinner
Race You There
You have to sign a lot of paperwork to get brain surgery. You— and your family— need to sign a lot more if you’re only fifteen.
“And, there’s always the risk of internal restructuring at this age…” Alexander isn’t listening to the doctors. He’s picking at the skin of his finger nails and resisting the urge to roll his wheelchair out of the room. This is the third and final specialist they need to sign-off on the procedure, and she’s not saying anything he hasn’t heard before. You know, the brain may be finished growing in size at your age, but it’s still growing in other ways… Alex knows that this is important information, and he did seriously worry about it a year ago when the process started. But now he’s accepted the risks. More importantly, he’s ready for the rewards. If it becomes dislodged due to restructuring of the motor cortex, we’ll have to wait another five years to try again…
More forms are signed. They could wait a few years, and boost the chance of long-lasting success from 70% to 90%. But together, Alexander’s family has decided that it’s worth a higher chance of failure if it will give Alex the rest of his young adulthood with the ability to walk. He’s only going to be a teenager once; he wants to play soccer with his friends after school and navigate college campuses without needing to map his route beforehand to ensure accessibility. The last doctor signs the last form. The next month, Alex is being prepped for brain surgery.
As he sleeps, a team of neurosurgeons and engineers hover around the bed. They cut through the skull and slide a machine the size of a penny into the soft, squishy organ they reveal underneath. Then, with the help of robotic surgery-aids that never shake or wobble, they arrange a net of hair-like fibers over the motor cortex. It looks like a spider has been at work in Alex’s brain. The engineers conduct tests as everything falls into place; they double and then triple check that everything is working as expected. There are no second chances to properly calibrate this tiny machine. Wireless connection established… protections against outside interference in place… every node is reading the associated neuron correctly…
When Alex wakes up, they tell him to rest. He doesn’t want to rest— he gets to practicing as soon as they will let him. At first, they tell him just to move a cursor on a computer screen. It takes two days of calibration, frustration, and headaches before he can do it. But from there, progress is faster. He can play chess with the doctors using just his thoughts. He can navigate his electric wheelchair around the hospital just by rolling his shoulders and imagining using his arms. Three weeks later, the doctors say we think you’re ready for the braces, and he almost cries for joy. Almost.
Alexander’s legs are fitted with custom braces. They can be expanded as he grows, accounting both for height and, hopefully, new muscle mass. Its mechanized joints make him feel a bit like a marionette— but no one else’s hands will move him. When the doctors connect his implant to the brace, it’s Alex’s own thoughts that cause the joints to move. Sporadic at first, he can bend his knees a few degrees, turn his ankle, sometimes even lift a leg with intention. A physical therapist teaches him how to walk alongside a technician who teaches him how best to concentrate on the idea of walking, how to sharpen the commands he sends his body so that the implant and the brace and his body can function seamlessly together. After three days, he can get out of bed and stand on his own. After a week, he can walk.
“Race you there!” he calls to his bewildered family as he runs through the corridors of the hospital a month after his surgery, relying on nothing but a leg-brace and his own mind.
A Biological Computer
The human brain is often likened to a computer for its ability to store and process tremendous amounts of complex information. In many ways, this is an apt comparison. Rhythmic electric patterns known as “brain oscillations” help to store, retrieve, and regulate information flow between neural circuits. Electrical synapses are therefore critical for human consciousness, a mechanism of computation that is “fundamentally analog”. However, the brain also displays “digital” processes, such as the history of histone and epigenetic methylation (modifications to DNA and gene expression caused by biological triggers) having a measurable impact on memory storage. In fact, the brain seems to rely on the transition of one-dimensional, “digital” information contained in DNA to three-dimensional, “analog” information stored in proteins. This process is unique to biological systems and is one of many reasons why “artificial intelligence”, even those with seemingly lifelike conversational skills, is unable to truly simulate human thought. Thus, rather than try to replace human brains with AI, some innovators are looking to instead augment our existing minds with computers.
The field of brain-computer interfaces (BCIs) is not all new and speculative. In 1924, German psychiatrist Hans Berger successfully recorded electrical brain signals from a patient. His discovery led to the development of electroencephalogram (EEG) tests, a method which, by attaching nodes to the scalp, a machine measures and graphs cerebral electrical activity. EEGs are used in the medical field today in the treatment of epilepsy, traumatic brain injuries, and sleep disorders. In 1988, an EEG interface allowed a user to “mentally” stop and start a line-following robot by measuring alpha rhythms in the brain, which were altered by opening and closing the eyes. In 2009, Mattel released Mindflex, a game which allowed the player to control the movement of a ping pong ball by “concentrating” while wearing an EEG headset. Critics have reasonably questioned if the Mindflex device is not moving the ball at random while relying on illusion of control cognitive bias to amuse users; independent experiments suggest that perhaps half of users demonstrated an actual ability to manipulate the ball’s movement by changing their mental state. In 2020, NextMind released a wearable EEG headset and “dev-kit” which was intended to facilitate hands-free software interaction. By visually concentrating on a button or other interfaceable element, the sensors interpreted associated brain activity and mapped it to an action in the software. Snap Inc., the company which acquired NextMind in 2022, stated its intended effect was to allow users to “push a virtual button” with mental focus, not read thoughts or discover new commands which had not been properly mapped. The technology proved finicky and difficult to use; Snap ultimately did not continue with the development of the project.
Modern Upgrades
Of course, BCIs are not going to be limited to EEG headsets forever. The world of brain-computer interfacing is rapidly accelerating and changing shape. Perhaps the most well-known (and thus far most successful) innovator in this dynamic field is Neuralink, a company founded by tech mogul Elon Musk. Neuralink isn’t using scalp-nodes and EEG headsets; instead, they are pioneering a more invasive form of BCIs which many colloquially call “brainchips”. The N1 Implant is a coin-sized device with microscopic electrode-threads which is surgically situated in a target section of the brain. When neurons fire in that area, the signal is recorded and transmitted to an external computer. This invasive method doesn’t just allow for faster and more complex signals from the brain to a machine, it also hypothetically allows the implant to transfer data back into the brain using the web of electrode cables, something which cannot be replicated by EEG technology.
Medical Miracles
Neuralink conducted a series of controversial primate trials before finally releasing proof of their success in the form of a viral video wherein Pager, a rhesus macaque with an N1 Implant, played Pong with only his mind. In 2023, the company was granted FDA approval to begin human trials. Neuralink advertised its experimental trial “PRIME” to individuals who had limited or no ability to use their hands due to spinal injury or ALS. By January 2024, they had successfully operated on their first patient: Noland, a man living with quadriplegia for eight years. Within two weeks, Noland was able to connect to a computer and play online chess, browse the Internet, and even use a keyboard to send messages by guiding a cursor to each letter individually. RJ, another PRIME participant who received his implant in April 2025, is able to operate both his computer and smartphone. Both men espoused praise for the care they received from Neuralink’s team and the success of the program, which gave them a new means of online and digital freedom. But Neuralink believes their N1 implant can do even more to give disabled people autonomy. In November 2024, the company launched CONVOY, a feasibility study examining whether PRIME participants could use their implant to control machinery, such as a robotic limb or a wheelchair. Like its sister-study, CONVOY has seen success: PRIME study member Alex is also the first participant in CONVOY to move a robotic arm with only his mind, allowing him to switch lights, open or close doors, and move objects obstructing his wheelchair.
The implications for CONVOY in the field of bionics and neuroprosthetics are tremendous. If those with quadriplegia (or any other form of paralysis or neurological disease) are able to control an external limb, they may also be able to control a sophisticated limb-brace or mechanical movement aid attached to their bodies. As Snap Inc. said, BICs are not mind-readers; rather, they are interpreting brain signals which command a limb to move a certain way (in PRIME’s case, the N1 implant is translating signals for hand movement to a cursor which is used to interact with a digital environment.) It is therefore feasible that the interface could be taught to, upon receiving a signal to “lift the leg” or “step”, move a series of mechanical joints to perform the associated movement on the wearer’s leg, thus allowing those with paralysis to walk again. For amputees, an implant-controlled robotic limb can be worn in place of the missing limb. This form of neuroprosthetic has already seen some success in clinical trials through EEGs which detect peripheral nerve signals (PNS). By creating a more direct interface between the brain and prosthetic, significantly more finesse could be granted over the limb. Robotic fingers could someday be precise enough to allow musicians and athletes to play again, artists to create, or professionals to work with fine instruments.
Of course, this is more complicated in application than theory. While an EEG headset can detect electrical activity from all sections of the brain, a Neuralink implant must be surgically placed and wired into a specific region. This allows it to have extremely precise readings, but also limits its scope. A single implant may only be able to control a single limb, or only the limbs on one side of the body. As of today, there are no approved clinical trials for multiple implants, and the potential dangers of repeated surgeries and physical modifications to the brain are greatly compounded compared to a singular device.
But sending mental signals to computers isn’t all that Neuralink hopes to achieve; their product Blindsight aims to use computers to map signals back onto the brain itself. While some vision-loss therapies try to repair or regrow pieces of the eye, Blightsight circumvents the entire optic nerve interface entirely. Instead, this hypothetical implant uses electricity to stimulate a neuron in the brain known to correspond to a specific point in a subject’s field of vision. They then “see” a phosphene— a bright spot in vision that was not caused by light entering the eye. Through mass coordination of these electrical stimuli, the brain can be fed sufficient information to perceive shapes, colors, and hopefully complete images, with resolution being dependent on the concentration of neurons being targeted. At the current stage of technology, Neuralink doesn’t think they’ll be able to restore complete sight; rather, through connection to a camera worn on or near the head, Blightsight may be able to show a grid of black and white dots which the brain, being incredibly sophisticated at pattern-recognition, eventually learns to interpret as a navigable environment and meaningful images. Further iterations, with deeper threading between the N1 Implant and the brain along with more complex stimulating technology, may be able to grant a complete simulacra of human vision. After successful primate trials in which a vision-impaired monkey was able to see and navigate a computer-generated maze, Blindsight received “Breakthrough Device Designation” by FDA, with expectations that human trials would begin in 2026. As of today, no formal trials are scheduled, but both Musk and Neuralink seem confident that Blindsight will advance within a matter of months.
Fun and Brain-Games
Just as EEGs were created as a life-saving medical tool and advanced to the point of allowing children to play a ball game with their minds, there’s little doubt that brain implants of the level of sophistication as the N1 Implant will have an array of recreational uses— for those who can afford the financial investment and medical risk. You wouldn’t need a headset to play pong; instead, you could command a robotic companion or control a drone using just hand or eye movements as an interface between you and the machine. You could lazily browse the internet on your phone or laptop without needing any physical contact to the device. Programs like CONVOY could be used to give people without disabilities additional artificial limbs to play or work with as they please, or to modify the ones they have (for example, robotic braces which make you appear taller but move as fluidly as your own legs.)
If Blindsight proves successful, the recreational implications grow even more tremendous. Perhaps you would not even need a device to access the internet— the N1 Implant could stimulate your brain to perceive a “holographic” depiction of a screen in your field of vision, and then assess your corresponding neural signals to allow you to scroll, click, and type. Elon Musk himself has even claimed that Blindsight may someday give the user “superhuman” sight, such as seeing into the infrared spectrum. Furthermore, if the N1 Implant can stimulate optical neurons, perhaps it could also be used to stimulate auditory or tactile regions of the brain. You could listen to music inside of your head, watch a movie and experience “4D sensations” like wind, rain, and heat on your skin with no external stimuli, or receive tactile feedback from “your” fingers while operating an artificial limb. These ideas are only a fraction of what would be possible with human ingenuity and such rapidly-evolving technology in the age of machine-learning. Today, Neuralink is only experimental medical technology in its infancy; in the next decade, who knows what novel and unforeseen uses will have been invented for invasive BCIs?
Issues of Safety
Although PRIME and CONVOY trials have proven the early success of the Neuralink implant, invasively-situated devices carry an array of risks— not to mention the general dangers of conducting surgery on the brain. Repeated surgeries to fix mechanical failures or add chips to communicate with additional regions of the brain can add exponentially to the risk of serious injury. Noland, the successful PRIME trial patient discussed earlier, experienced this firsthand. Only a month after the implantation, Noland’s chip stopped working: the Neuralink team said they hadn’t accounted for how much the human brain “pulses” with our heartbeats, and over 85% of the wiring had come undone. Despite Noland’s insistence that they operate again to reattach the fibers, the team said they were wary of attempting another invasive surgery. Instead, they reprogrammed the remaining 15% of the neural net to read from groups of neurons, rather than individual cells. This gave him control of the implant again, but cast uncertainty on the overall longevity of the device. Were the remaining threads to come unseated from his brain, he and his care team would be faced with the difficult choice between risking permanent brain damage in an effort to either replace or realign the implant (with no guarantee that the movement of the brain won’t dislodge it again) or to do nothing and let Noland lose his autonomy once again. Were a member of the Blindsight trial to experience a “slippage” of the neural network in this way, the results could be much worse than simply losing contact with the implant: electricity delivered to random neurons in the brain could not only cause random and disorientating phosphenes, but also muscle spasms, migraines, and even seizures. Similar instances of misfiring could also result from glitches in the software or from the implant decreasing in reliability over time. As of now, there is no known “lifespan” for a Neuralink chip— even if it were perfectly situated in the brain for years, the physical components themselves would eventually degrade in functionality. Would this decay necessitate removing the implant to avoid damage to the brain? How many times in a human lifetime could an implant safely be replaced?
Finally, as with any technology that uses wireless communication, there are concerns about hacking. Although hackers face a number of hurdles— especially if the implants themselves are not connected to the internet— cybersecurity expert Roger Grimes believes that hacking is a very real, very dangerous risk for Neuralink chips. “Biohacking” medical devices is not purely hypothetical; it’s been happening for over a decade: Vice President Dick Chaney’s defibrillator had to have its wireless features modified after hacking attempts, the FDA recalled almost 500,000 Pacemakers after discovering they could be hacked, and Johnson & Johnson warned patients that their insulin pumps were vulnerable to outside interference. Just as hackers can threaten heart attacks and insulin overdoses in today’s world, future “brain-hackers” could access attached devices like computers (stealing patient data, identities, etc), wheelchairs, or limbs, causing serious physical harm to the user and those around them. If the chip sends data back into the brain, hackers could cause auditory or visual hallucinations, or even kill the user by causing severe seizures.
Neuralink— and any other brain-implant brands or innovations which will no doubt follow— have the potential to revolutionize quality of life for those living with disabilities. It may someday be able to grant complete mobility and sight to those who have lost theirs, and perhaps even become one of the most expensive new toys on the market. In getting there, we must ensure that, alongside these exciting developments, proper care is taken to insulate invasive BCIs from biohackers and that the risks to the brain of prolonged use, multi-use, and repairs are completely understood. To push advancements forward without necessary precautions could only further destroy the lives we are trying to uplift. But hopefully, with all eyes on Neuralink and some of the greatest biomedical and tech minds working together to address these concerns, we will soon live to see another great leap forward in healthcare and human wellness.
