Our recent focus on robotics and artificial intelligence in the food service industry has left us hungry to investigate parallel developments in other spheres. The bio-medical field, for instance, is one area that narrows the distance – both theoretical and physical – between conceptual artificial intelligence and its physical implementation. According to the Food and Drug Administration (FDA), medical devices (instruments that affect the body’s physical structure or function, independent of chemical actions or the metabolism of pharmaceuticals) fall into one of three classes: Class I, for example bandages, is the least risky; and Class III (for instance, pacemakers) represents the greatest risk. So it’s no surprise that the experimental device we’re exploring today, currently in development by Elon Musk’s San Francisco, Calif-based Neuralink, could be considered a Class III device, on steroids. Regarded as a ‘Breakthrough Medical Device’ per the FDA, Neuralink’s product is considered to potentially satisfy two criteria: it will address a ‘life-threatening or irreversibly debilitating’ condition; and it will provide new or unique advantages over existing technology.(1) And why is this important? Because, in terms of FDA expedition of reviews and possible approval, different rules rules apply to breakthrough devices. And this is concerning as we begin to wrap our minds around the nature of the technology involved. In fact, it is possible that the very sovereignty of our minds is the most serious consideration of all in this work. How? Let’s find out more…
Last week, innovator and entrepreneur Elon Musk took to the virtual stage to present a product update via the YouTube platform.
Musk, who is known primarily for SpaceX, Tesla semi-autonomous vehicles, and the fact that he has – according to Forbes magazine – just become the world’s fifth $100 Billion-Dollar man, headlined the event to generate enthusiasm for what the innovation titan is describing as a ‘Fitbit in your skull.’(2,3) Offering users the opportunity to ‘[r]ock-climb without fear. Play a symphony in your head. See radar with superhuman vision. Discover the nature of consciousness. Cure blindness, paralysis, deafness, and mental illness,’ the device is a brain-computer interface (BCI), an implant that Musk forecasts will forge new neural pathways that revolutionize an individual’s thought processes.(4) A significant claim indeed. So what does this system look like and what kind of technology is involved?
Also termed brain-machine interfaces (BMIs), the group of devices that function as a link between the human brain and the computer are stimulating a great deal of excitement with their promise of restoring motor function, sensory input, the ability to treat neurological disorders, and their potential in the treatment of bio-psycho-social conditions such as depression and addiction. Since its founding in 2016, Neuralink has been famously reticent about discussing its research but, at last week’s update, more than 100,000 viewers got to witness a snapshot of its progress. Resembling a sewing machine bobbin, what Musk is calling ‘the link’ is an implant approximately the size of a silver dollar coin with super-thin electrode threads. Approximately the thickness of the human skull, the link is designed to fit on top of the brain in a hole drilled into the skull.
Yes, you did read that correctly: following the removal of a section of bone, the implant is attached directly into the brain via electrodes.
Finer than a human hair, the electrode probes are engineered from biocompatible materials – a polyimide substrate and dielectric with encapsulated gold thin film. The thin film arrays comprise two distinct sectors: a thread area with electrode contacts; and a sensor area to interface with chips that capture, amplify, and wirelessly transmit neural signals. In terms of power and up-time, Musk sees the device as being charged wirelessly by an induction coil that users would plug in overnight to recharge, however the details of that aspect of the project remain sketchy at this time.
What we do know, however, is that portions of the work on the platform have been conducted by a number of California-based institutions. These include Lawrence Livermore National Laboratory (LLNL) which offers a 4000 sq ft Class 100 cleanroom with a dedicated medical device foundry. The laboratory is equipped to engineer devices from advanced prototypes through human clinical studies and boasts the ability to offer deposition tools for polymers, metals, and dielectrics, integration and assembly tools for electronics and encapsulation, and metrology tools for use in ellipsometry. Furthermore, a characterization lab is also available ‘for neural interface characterization; electrode and sensor development; and electrical, chemical, and mechanical lifetime testing.’(5)
Facilities at UC Berkeley seem to be equally as impressive, hosting not only the Berkeley Wireless Research Center which is contributing to Neuralinks’s project, but also the Berkeley Artificial Intelligence Research Lab, the Laboratory for Automation Science and Engineering, the Center for Automation and Learning for Medical Robotics, and the Center for Human Compatible Artificial Intelligence, among others. At these centers researchers have access to the Marvell NanoLab which offers in excess of 15,000 sq feet of Class100 and Class1000 cleanroom as a shared resource with all necessary micro- and nano-fabrication tools.
Furthermore at another Neuralink partner, Stanford Nanofabrication Facility, device fabrication resources are offered within a 10,000 sq ft cleanroom and two labs. These satellite facilities support Metallo Organic Chemical Vapor Deposition and pioneer experimental fabrication methods.
All in all, Neuralink seems to have access to some of the best resources – both human and mechanical – that the nation has to offer. But even with this level of investment in research and technology, two flaws inherent to computing remain inescapable: deprecation and obsolescence. According to an article in MIT Technology Review, the implant has been carefully designed to be easy to install and to remove ‘so that people can get new ones as technology improves. You wouldn’t want to be stuck with version 1.0 of a brain implant forever.’(6) Heavens no, a brain implant v1.0 is so last season…
Moreover, at this point in development, Neuralink is downplaying risks of the surgery, claiming that the procedure should take only around 30 minutes and should require no general anesthesia. Plus the entire implant surgery will be performed by a robot.
According to an article by Musk in preprint, ‘An integrated brain-machine interface platform with thousands of channels,’ the design of the robot insertion device offers a solution to one of the most intractable problems of using thin film polymers for electrode probes: stiffness. Deploying the robot allows for the variable speed insertion of a large number of flexible electrodes which are sited such that they avoid blood vessels but still penetrate both the meninges and the underlying brain tissue. How? ‘The robot registers insertion sites to a common coordinate frame with landmarks on the skull, which, when combined with depth tracking, enables precise targeting of anatomically defined brain structures.’(7) In other words, the robot learns the physiological landscape of the head and the brain by recording its landmarks. Thereafter, custom software ‘allows pre-selection of all insertion sites, enabling planning of insertion paths optimized to minimize tangling and strain on the threads. The planning feature highlights the ability to avoid vasculature during insertions, one of the key advantages of inserting electrodes individually. This is particularly important, since damage to the blood-brain barrier is thought to play a key role in the brain’s inflammatory response to foreign objects.’(8)
And this last point is one that gave us pause for thought in terms of patient safety and risk for contamination. The robot is being designed by Woke, a Canadian company that specializes in designing for technology. Included among its clients are companies as diverse as Intel, Hyundai, Amazon, and Lululemon among others, all of which share one common concern: the creation of product designs that connect people with technology. Since Woke has been at the forefront of designing Neuralink’s robotic installer, this arguably the zenith of projects connecting people with ‘technology in a very powerful way.’(9)
And this is terrific, assuming that the robot can be scrupulously decontaminated between procedures, as we would hope it to be. However, let’s not forget that we have seen other cases in which medical instruments and devices have been the causes of infection. Do you recall the article we brought you on the reprocessed duodenoscopes? If not, here’s the link. And of course, on a slightly different note, there was that tragic failure of PPE during the ebola crisis in Sierra Leone and Ivory Coast in which ‘strike-through’ led to the deaths of medical personnel. So it’s definitely not without precedent that bio-medical equipment can fail both in its function and in its cleanliness. And in the case of an implant attached directly to the surface of the brain we can only surmise the scale of the potential risk if contamination is present. In the case of catheters used in angioplasty, for instance, we know from Berkshire’s own Technical Brief that ‘an FDA guidance document recommends measurement of the total quantity and size(s) of the particulates generated during the simulated use of the device. This recommendation is based on the premise that “if particles are introduced in the bloodstream during an angioplasty procedure, they may present an embolic risk to the patient”.(10) Furthermore, in any surgery, the document continues, ‘measurement of the total quantity and size of particulates a device may generate is an indication of embolic risk [because] particulate escaping into the vasculature can ultimately accumulate in the brain, occluding small vessels and leading to ischemic stroke”.(11)
So, given that surgery on other parts of the body holds a potential for contamination to affect the brain, how much greater is the risk when the brain itself is the locus of the intervention? Let’s return to the Technical Brief one last time: ‘Particulate contamination will take on even more significance as technological advancements occur with the use of robotics and nanotechnology which focus on the reduction of scale and particulate size. It will be incumbent on manufacturers to understand particles and develop the proper procedures to minimize contamination on the devices they create.’(12)
The danger is real and the onus is upon researchers and manufacturers to provide guardrails against contamination, in just the same way as they do in the fields of existing bio-medical device manufacturing and of transplant surgery.
Our established protocols that ensure patient safety should form the foundation of an extended suite of measures that minimize the hazards inherent in this endeavor. That is, assuming of course that the technology is ethically sanctioned and socially accepted – a mighty assumption that we should not necessarily make at this time.
Since this is a technology still very much in experimental phases, it’s impossible to wrap up our discussion neatly and throw a spotlight on its future. However, it is interesting, at least, to examine the ways in which the question of human enhancement via a direct connection to technology is not a new one. In popular culture, for instance, works of both science fiction and speculative fiction have long mined a rich narrative seam at the intersection of robotics, AI, and the human experience. From British author Iain M. Banks’ Culture novels to the cybermen of the iconic TV series Dr Who, writers have explored the ways in which the combination of human intellect and computer intelligence could be leveraged for good or ill in pushing the boundaries of possibility.
Furthermore, as far back as the early 1950s when evolutionary biologist and eugenicist Julian Huxley first coined the term, what would become the transhumanist movement has welcomed advances in robotics, AI, brain and body modification, among others, in the pursuit of a technofuture itself rooted in sci-fi. So it is interesting that when we scratch the surface of the official, quasi-medical narrative, we see a glimmer that Musk’s motivation for pioneering this BCI implant may indeed lie more in an admiration of transhumanist precepts than in a drive to cure paralysis or addiction. Indeed according to the MIT Technology Review, Musk’s presentation ‘continually drifted away from medicine and back to a much more futuristic “general population device,” which he called the company’s “overall” aim. He believes that people should connect directly to computers in order to keep pace with artificial intelligence.’(14) Moreover, it is no secret that Musk has been greatly inspired by the Banksian concept of ‘neural lace,’ an all-purpose implant that acts as a BCI, improves memory, enhances biological functions, and aids in communication, and therefore unsurprising that Musk may be most strongly motivated by a more direct connection with AI.
Why? That’s an excellent question and, superficially at least, a conundrum.
Since reading Nick Bostrom’s work ‘Superintelligence: Paths, Dangers, Strategies’ in 2014, Musk has gone on the record warning of the potential dangers of our race to create artificially intelligent entities: ‘I have exposure to the very most cutting-edge AI and I think people should be really concerned about it.’(12) Equally: ‘I think we should be very careful about artificial intelligence. If I were to guess like what our biggest existential threat is, it’s probably that.’(15) Despite these reservations and the fact that Neuralink’s research has thus far been confined to early animal trials with rodents and current ones using pigs as test subjects, the intention is clearly to move onto the recruitment of human volunteers. Per Musk’s preprint: ‘it is a research platform for use in rodents and serves as a prototype for future human clinical implants.’(16) And how Neuralink sells the idea to volunteers and future users alike is a question for our time. Direct, unspoken mind-to-mind communication, the ability to control smart devices by thought alone, and the ability to back up a life’s worth of memories for a potential restore or download into a robot body have all been suggested as applications of the technology.
And, as such, they may sound tempting to volunteers who, in any technofuture, will undoubtedly be revered as pioneers. However, as Musk also commented, “On a species level, it’s important to figure out how we coexist with advanced AI, achieving some AI symbiosis […] such that the future of world is controlled by the combined will of the people of the earth. That might be the most important thing that a device like this achieves.”(17)
Or it might be that Musk and his ilk have seriously underestimated the ‘existential threat’ posed by AI and our flirtation with technological enhancement, such that the race to keep pace with artificial intelligence is one that we might ultimately lose. And we have to wonder what the landscape of that particular brave new world would look like…
Transhumanism and technologically-enhanced bodies – are you intrigued by the possibilities? Or do research projects such as Neuralink’s inspire more fear than excitement? We’d love to know your thoughts.