AI Library
The Age of Autonomous Scientific Discovery
Kim Kyung-jin, Attorney at Law
AI Scientists and Self-Driving Labs
This book follows how AI scientists and self-driving labs are changing the way science generates and verifies claims. It covers literature-based discovery, natural-language protocols translated into robot commands, multi-agent research systems, closed-loop laboratories, materials search, the verification gap, chains of evidence, research harnesses, journal ethics, and legal responsibility.
AI Library
A New Era of Life Sciences Opened by Artificial Intelligence
Structural Proteomics, Genomic Foundation Models, Autonomous Laboratories, and Global Governance
Kim Kyung-jin, Attorney at Law
This book is a research volume compiled with artificial intelligence. A human selected the materials and structured the work, while AI models drafted the sentences and cross-checked the facts.
AI Library
The Double Structure of Digital Sovereignty
Europe’s Departure from Palantir and the Chains of American Big Tech
Kim Kyung-jin, Attorney at Law
This is a record of 2026, when European intelligence agencies and defense ministries began removing analytics tools from America’s Palantir. It covers the replacement decisions made by France’s General Directorate for Internal Security (DGSI), Germany’s Federal Office for the Protection of the Constitution (BfV), and the Netherlands Ministry of Defense; the incident in which US export controls severed an ally’s ac…
New English Edition
Artificial Intelligence in Horticulture
Kim Kyung-jin, Attorney at Law
Across five chapters and ten sections, this book examines computer vision for crop diagnosis, harvesting robots and autonomous field systems, smart greenhouses and digital twins, precision irrigation and supply-chain quality control, high-throughput phenotyping, and predictive breeding.
New English Edition
Artificial Intelligence in Food Crop Agriculture
Kim Kyung-jin, Attorney at Law
Across six chapters and eighteen sections, the book examines digital agricultural infrastructure, remote sensing, crop diagnosis, yield forecasting, precision irrigation, genomics, molecular breeding, agricultural robotics, climate-smart agriculture, and global food security.
New English Edition
The Future of Forestry and Agroforestry
Kim Kyung-jin, Attorney at Law
Driven by Artificial Intelligence and Digital Innovation
Across five chapters and fifteen sections, the book follows satellites, drones, LiDAR, digital twins, forest-specific language models, wildfire and pest forecasting, forestry robotics, agroforestry, timber traceability, and forest carbon markets.
New English Edition
Smart Livestock Farming: AI Enters the Barn
Kim Kyung-jin, Attorney at Law
Sensors listen, cameras watch, and artificial intelligence helps farmers decide.
Across five chapters and fifteen sections, the book follows precision livestock farming from animal health and reproduction to robotic milking, virtual fencing, digital twins, methane reduction, welfare, and data ownership.
Table of Contents
Han Dong-hoon, Busan Buk-gu Gap: A Record of the 100 Days Before and After the Election (Mar. 26-Jul. 3, 2026)
Kim Kyung-jin
Table of Contents and 13 sections
From March 26 to July 3, 2026, this record follows the spring after expulsion, the Busan Buk-gu Gap by-election, victory as an independent, and the first bill submitted in the National Assembly.
Table of Contents
Artificial Intelligence and Medicine
Kim Kyung-jin, Attorney at Law
AI in clinical care, hospitals, education, and research
AI in medical imaging, risk prediction, treatment planning, hospital operations, education, and research, with patient safety, privacy, and accountability.
[AI Library] Chapter 4. The Birth and Vision of Neuralink
Brain Readers: Neuralink and the Final Human Revolution
Chapter 4. The Birth and Vision of Neuralink
Kim Kyung-jin
Part 2. Elon Musk's Gamble: Neuralink's Experiments and Breakthroughs
A. 2016: The Secretive Founding and Its Key Players
One afternoon in October 2016, a young engineer named DJ Seo picked up a phone call in a lab at UC Berkeley. A doctoral student at the time, he was researching ultra-small sensors called "Neural Dust." The voice on the other end was familiar. It was Elon Musk.
Musk got straight to the point. "Don't you want to commercialize your research?"
Seo didn't hesitate. He decided to join on the spot. On his first day, he arrived at a small office in San Francisco's Mission District, and what greeted him was an empty room. There weren't even chairs. He had to drive to Office Depot and buy one himself. That's how Neuralink began.
Neuralink was officially registered in California in July 2016. But the public didn't learn of its existence until March 2017, when the Wall Street Journal broke the story. For roughly a year, the company operated in stealth mode, quietly recruiting talent across Silicon Valley and laying its technical foundations.
The founding team was assembled by design. Musk interviewed more than a thousand candidates. He wasn't just looking for brilliant scientists. He was searching for people who could fit together puzzle pieces from neuroscience, robotics, semiconductors, and materials engineering.
Eight co-founders came together.
Max Hodak was an entrepreneur who had studied biomedical engineering at Duke University. As an undergraduate, he worked in the lab of Professor Miguel Nicolelis, where he helped build brain-machine interfaces tested on monkeys. After graduating, he founded Transcriptic, a cloud robotics lab platform. He became Neuralink's first president, overseeing day-to-day operations.
Benjamin Rapoport was both a neurosurgeon and an electrical engineering PhD. He was a rare find, someone who could bridge the gap between the realities of the operating room and the ideals of engineering. He designed the early clinical protocols for implanting devices safely in the brain while minimizing surgical risk.
DJ Seo's Neural Dust research focused on ultra-small wireless sensors that communicate with the brain using ultrasound. This work became a technical inspiration for Neuralink's philosophy of minimal invasiveness. Seo would later rise to become the company's president and CEO.
Paul Merolla had designed neuromorphic chips at IBM, chips modeled after the brain itself. He took charge of the semiconductor core, building processors that could handle massive volumes of neural signals on minimal power.
Philip Sabes was a professor at UC San Francisco and an authority on motor control neuroscience. His career had been spent studying how the brain sends commands to muscles and how those signals could be decoded.
Tim Gardner was a neuroscientist at Boston University who studied birdsong. His work tracing how complex vocal patterns emerge from a small bird's brain connected directly to Neuralink's long-term goal of restoring human speech.
Tim Hanson was an expert in robotics and precision machinery. He designed flexible electrodes to replace the rigid ones used before, and he developed the early concepts for a robotic system to insert them into the brain.
Vanessa Tolosa, a neural engineer from Lawrence Livermore National Laboratory, took on one of the hardest problems: developing materials that could stay inside the brain for years without corroding or triggering an immune rejection.
This combination of eight people was no accident. Musk understood that a problem like brain-computer interfaces could not be solved by any single discipline. Neuroscientists alone weren't enough. You needed semiconductor engineers to design the chips. You needed surgical robotics specialists to put those chips into the brain. You needed materials scientists to create substances the brain wouldn't reject. Every piece had to interlock for the puzzle to be complete.
But cracks began to appear even in this seemingly perfect team.
Musk demanded Silicon Valley speed from researchers who came from academia. His favorite phrase was "maniacal sense of urgency." He pushed them to build working products right now, not write papers. The collision between intellectual curiosity and commercial pressure was inevitable.
Co-founder Rapoport left the company in 2018. He cited safety concerns as a primary reason for his departure. He then founded a competitor called Precision Neuroscience,
which chose to place a thin film on the surface of the brain rather than using Neuralink's penetrating electrodes.
Hodak left in 2021 as well. He posted briefly on Twitter that he had "not been at Neuralink for a few weeks," without explaining why. He, too, started his own neurotechnology company, called Science Corp.
As of January 2022, only two of the eight co-founders remained at the company: Musk and DJ Seo.
A 2020 report by STAT News described the internal situation this way: "Neuralink has been through years of internal turmoil. Conflicts have persisted between a rushed timeline and the slow, incremental pace of science."
Still, Neuralink kept moving. The company that started in an empty office had grown to 90 employees and $158 million in funding by 2019. That year, Neuralink revealed its technology to the public for the first time. Alongside a white paper, the team presented prototypes of flexible electrodes and a surgical robot. They announced they had successfully read signals from a rat's brain using 1,500 electrodes.
Those three years behind the veil were not wasted. Without that quiet preparation, implanting a chip into a human skull in 2024 would have been impossible.
B. Musk's Goal: Human-AI Symbiosis as a Defense Against the Threat of AI
To understand why Elon Musk founded Neuralink, you first have to understand his fear.
In 2014, Musk warned in an interview. "Artificial intelligence could be more dangerous than nuclear weapons."
His concern was not idle speculation. Google's DeepMind had defeated the world Go champion with AlphaGo. AI performance in image recognition, speech recognition, and natural language processing was improving exponentially every year. Musk imagined what would happen if this trend continued.
His analogy was blunt. "If AI becomes far smarter than humans, we'll be like house cats to AI. Cute but helpless."
There were worse scenarios too. AI could ignore and crush humans the way we ignore ants. To avoid this dystopia, Musk put forward a single piece of logic.
"If you can't beat them, join them."
This is the philosophical foundation of Neuralink. Symbiosis between humans and AI. For Musk, Neuralink was never just a medical device company. It was a survival tool for humanity to make it through the age of AI.
Musk reinterpreted the human brain from a hardware perspective.
The first layer is the limbic system. It is the oldest part of the brain, responsible for survival instincts, drives, and emotions. The second layer is the cortex. It handles logical thinking, planning, and language. It is the layer that makes humans human.
Musk argued that a third layer already exists. He called it the "digital tertiary layer."
Think about your smartphone. We can already access all of humanity's knowledge through Google. We can communicate with anyone in the world via email and messaging. Complex math is hard to do without a calculator. In a sense, the smartphone is an extension of our brain.
Remember the anxiety you feel when you leave your phone at home. That sense of loss, as if a part of your body is missing. Musk said this was the proof. Modern humans are already a kind of cyborg.
But there is a problem.
The connection between the digital layer and the biological brain is too slow. AI computes at trillions of bits per second. Humans, at best, tap a screen with two thumbs or speak with their voice. Musk did the math. The human information output speed is only a few bits per second.
At a Y Combinator event in 2024, Musk said this. "There are 86,400 seconds in a day. It is extremely rare for a human to output more symbols than that number in a single day. The sustained output rate of a human is less than one bit per second."
This is the problem Musk defined. A problem of bandwidth.
AI communicates at terabit speeds. Humans communicate at bit speeds. Because of this enormous gap, humans cannot keep up with AI's decision-making speed. In Musk's words, "If AI is communicating at terabits and you're communicating at bits, it's like talking to a tree."
Neuralink's solution is direct. Build a high-bandwidth interface between the brain and computers. Let people communicate with the digital world at the speed of thought.
On Lex Fridman's podcast in 2024, Musk explained his long-term vision. "The long-term aspiration of Neuralink is to improve human-AI symbiosis. By increasing the bandwidth of communication."
What does that mean in concrete terms?
Musk introduced the concept of "consensual telepathy." Transmitting thoughts and concepts directly, without going through the imperfect medium of language. Suppose you imagine a blue elephant. Right now, you have to say "blue elephant." But with a high-bandwidth interface, you could transmit that visual image itself directly into another person's brain.
Uploading and downloading memories were mentioned too. At the extreme end, even the possibility that the mind could live forever in digital space after the body dies.
This vision is transhumanism at its peak. But the real Neuralink operates with far humbler goals. The reason is straightforward. Regulators and the medical field don't move for "human augmentation." They move for "clear medical benefit."
So Neuralink's first product, named "Telepathy," aims to let paralyzed patients control a computer with thought alone. "Blindsight" for restoring vision and "Convoy" for robotic arm control follow behind it.
The short-term goal is treatment. The long-term goal is augmentation. And the ultimate goal is symbiosis.
Critics call this dangerous techno-utopianism. But for Musk, Neuralink is insurance. Insurance for humanity to move into the future alongside AI.
Whether that insurance will actually work, nobody knows. But at least Musk has put more than $100 million of his own money on the line to answer that question.
C. Beyond the Bandwidth Limit: From 1-Bit Communication to Gigabit Transmission
Stephen Hawking spoke to the scientific community and the public for decades after ALS paralyzed his entire body. His tool was his cheek muscle. A sensor attached to his glasses frame detected tiny movements of his cheek and converted them into text. The speed was roughly one word per minute.
Musk said in an interview: "Enabling Stephen Hawking to speak faster than an auctioneer. That is the first goal."
That single sentence captures the technical challenge Neuralink has set for itself.
The problem is bandwidth. The speed of information transfer between human and machine.
The speed of typing on a keyboard, moving a mouse, issuing voice commands. All of these are absurdly slow compared to the speed at which the human brain thinks. Our input organs (eyes, ears) take in massive volumes of data every second. But our output organs (mouth, hands) are evolutionary bottlenecks.
In the history of brain-computer interfaces (BCI), 'one-bit communication' was a common starting point.
Yes or no selections. Spelling through brainwave responses to stimuli (P300). Limited commands like left, right, stop. These methods achieved reliability, but they were far too slow for real life. Communication at a rate of a few characters per minute is not enough to get through a day.
The standard for existing invasive BCIs was the Utah Array. This electrode, fitted with roughly 100 needles, could record signals from fewer than 100 neurons at a time. The human brain contains approximately 86 billion neurons. Trying to understand the brain through 100 channels is like trying to judge a full orchestra's performance by listening to a single violinist.
Neuralink raised that channel count by more than tenfold in one leap.
The N1 implant carries 1,024 electrodes. Sixty-four flexible threads each hold 16 electrodes. Each electrode samples neural signals 20,000 times per second. It can precisely analyze not only individual neuron spikes but also Local Field Potentials.
Why does this matter?
From an information theory standpoint, language is a very lossy compression algorithm. To convey a complex emotion or mental image to another person, we must compress it into imperfect symbols called 'words.' The listener then decompresses those words inside their own brain. Enormous amounts of information are lost in the process. Misunderstandings arise.
High-bandwidth connections change that story.
Take motor control alone. With 100 channels, you can roughly read the intention to 'reach out a hand.' With 1,000 channels, the possibility opens up to decode the pressure of pressing a piano key or the delicate tremor of a calligraphy brushstroke.
The same applies to restoring sensation. As bandwidth increases, not only reading but writing becomes more precise. Connect thousands of electrodes to the visual cortex, and you could transmit a low-resolution pixel image directly into a blind person's brain. That is exactly what 'Blindsight' aims to achieve.
Neuralink's first patient, Noland Arbaugh, moved a computer cursor with his thoughts after surgery in 2024. He played chess. He played Civilization VI for over eight hours. His brain signals were transmitted via Bluetooth to a computer, replacing mouse input.
Arbaugh described the experience: "It was like using the Force."
But 'gigabit transmission' remains closer to a slogan than reality. Actual clinical data shows rates of a few bits to a few dozen bits per second. The gap is enormous.
Why? There are constraints.
Power and heat. Brain tissue is sensitive to heat. Power consumption in an implanted device directly generates heat. You cannot warm the brain.
Wireless transmission. High-speed wireless communication consumes more power. Signals weaken as they pass through the skull and skin.
Decoding. Even if the number of channels grows, it is useless if algorithms cannot extract meaningful information. Pulling the intention to 'move the hand to the left' out of the raw data streaming from 1,024 electrodes is an entirely separate challenge.
Long-term stability. In Arbaugh's case, some electrode threads retracted from the brain within weeks of surgery. They pulled away from their original positions. The number of recording channels dropped. Neuralink restored much of the functionality through software adjustments, but the case left an important lesson. The bandwidth race is not a simple contest of electrode count. It is an all-out war encompassing mechanical fixation, tissue response, signal drift, and adaptive decoding.
Neuralink has a long-term roadmap to increase the electrode count beyond 16,000. Multi-chip implantation is also planned. Ten chips in one person's brain would mean 10,000 channels. The figure of one million channels has been mentioned.
But here and now, in the reality of 2025, Neuralink has enabled paralyzed patients to play chess, play games, and send emails with 1,024 channels. For some people, that alone has changed their lives.
The question of bandwidth is a question of human experience. How richly can a person connect with the world? And at the same time, it is also a question. When high bandwidth opens up, what happens to the private space of the human mind? In a world where thoughts are converted into digital signals, what does privacy mean?
D. Neuralink's Technical Differentiators and Barriers to Entry
In 2004, a research team at Brown University implanted an electrode called the Utah Array into the brain of a quadriplegic patient. The patient moved a computer cursor with thought alone. It was a world first. Twenty years have passed since then. Brain-computer interfaces are no longer a new concept.
So what sets Neuralink apart? As a latecomer, what has this company claimed as a competitive edge that rivals cannot easily match?
Three things sit at the core: flexible electrodes, a surgical robot, and a fully implantable wireless design. Binding all three is a vertical integration strategy that welds them into a single system.
First, flexible threads.
The conventional Utah Array is made of rigid silicon needles. Once driven into the brain, they stay locked in place. The problem is that the brain is soft. It has the consistency of tofu. It shifts with every heartbeat and breath. When a stiff needle sits inside soft tissue, each tiny movement lets the needle slice into the surrounding cells. Researchers call this 'micromotion' damage.
Over time the brain reacts to the foreign object. A glial scar forms around it. Once scar tissue encases the electrode, neural signals get blocked. Within months, or a few years at best, electrode performance degrades.
Neuralink took a different path. Its threads are made of polyimide, a flexible polymer. Each thread is 4 to 6 micrometers thick, less than one-tenth the width of a human hair (roughly 70 micrometers). These threads move with the brain tissue instead of fighting it. Tissue damage drops. The design is built for long-term signal stability.
Flexibility is an advantage, but it creates a new problem at the same time. The threads are so thin and soft that no human hand can implant them. It is harder than threading a strand of hair into a wet tissue.
That leads to the second element: the surgical robot, R1.
Neuralink built R1 in-house. Engineers call it the 'sewing machine robot.' A tungsten needle grips a flexible thread, pushes it into the brain, and then withdraws, leaving only the thread behind. It literally stitches the brain.
The robot's camera films the brain surface in real time. Tiny blood vessels spread across that surface like a spider's web. Nicking even one of them causes bleeding, and bleeding means brain damage. R1 uses computer vision to map the vessels and threads each electrode through the narrow gaps between them.
Speed matters too. The first-generation robot took 17 seconds per electrode. An improved version can insert 192 electrodes per minute. No human surgeon can match that precision or pace.
Musk has said the goal is to make the procedure 'as quick and simple as LASIK eye surgery.' The surgical robot is one of Neuralink's strongest competitive barriers.
Third, the fully implantable wireless design.
Think back to earlier BCI systems. BrainGate patients had thick cables protruding from their skulls, running to external computers. Infection risk was high. Normal daily life was impossible. The hardware only worked inside a lab.
Neuralink's N1 implant is different. It is a circular device about the size of a 500-won coin. Surgeons remove a small round section of skull and fit the device into the opening like a cap. Once the skin closes over it, nothing shows on the outside. Even the scar hides under hair.
Charging is wireless, using inductive power transfer through the skin. Data transmission is wireless too, relying on technology similar to Bluetooth. Musk described it as a 'Fitbit in your skull.'
Other companies can build each of these three technologies separately. Some make flexible electrodes. Some build surgical robots. Some design wireless implants. But Neuralink's real moat lies in vertical integration, fusing all of them into one seamless system.
Chip design, electrode manufacturing, surgical robot development, animal testing, clinical trials, and user app development all happen under one roof. Tesla builds its own batteries, motors, and software. SpaceX builds its own rocket engines, airframes, and launch systems. Neuralink follows the same playbook.
The payoff of this strategy is speed. When the electrode team finds a problem, it feeds back to the robot team immediately. When the robot improves, the change flows straight into clinical work. Feedback loops stay short. Dependence on outside partners stays low.
Competitors exist, of course.
Synchron does not open the skull at all. It delivers electrodes through blood vessels. The device, called the Stentrode, travels via catheter into a vessel near the brain, much like a cardiac stent. There is no craniotomy, no dread of having the skull cut open. That gives Synchron an edge in safety and public acceptance. Bill Gates and Jeff Bezos have invested.
Paradromics focuses on high bandwidth. Its Connexus system supports more than 1,600 channels and has reportedly achieved information transfer rates above 200 bits per second. In June 2025, a team at the University of Michigan completed the first human test.
Precision Neuroscience was founded by Benjamin Rapoport, a Neuralink co-founder. Its approach places an ultra-thin film called Layer 7 on the brain's surface. It is less invasive than penetrating electrodes. It is reversible; the film can be removed if needed.
Blackrock Neurotech is the original maker of the Utah Array. It has implanted devices in dozens of patients since 2004. In sheer volume of clinical data, Blackrock leads the field.
Neuralink's position in this competitive landscape is clear. More invasive, but higher performance. More risk, but wider possibilities. In the short term, the company starts with cursor control for paralyzed patients. In the long term, it aims at vision restoration, speech restoration, memory augmentation, and, ultimately, human-AI symbiosis.
The less-invasive competitors, meanwhile, bet on safety and acceptance to grow their user base faster.
Which side wins will not be settled by a technical debate alone. It will be decided by the combined weight of regulation, insurance coverage, adoption in clinical practice, and long-term safety records.
In January 2025, Neuralink's N1 implant system received 'de novo' classification from the U.S. FDA. It became the first fully implantable wireless BCI cleared for commercial use. The approved indication is external device control for patients with quadriplegia caused by spinal cord injury or ALS.
A moat is not built on technology alone. It includes the ability to navigate the distinctly human maze of law and regulation. Neuralink is still making its way through that maze.
Kim Kyung-jin
Attorney · Former Member of the National Assembly · AI Policy Researcher
© 2026 Kim Kyung-jin. All rights reserved.