On Wednesday night, Elon Musk hosted an update from brain-computer interface company Neuralink. Much of the update featured various researchers at the company providing an overview of the specific technology development areas they were working on. But there was nothing significantly new in the tech compared to last year’s update, and it was difficult to put the presentations together to create a cohesive picture of what the company planned to do with its hardware.
But probably the most striking thing was that last year’s update showed Neuralink approaching human testing. More than a year later, these tests are about six months away, according to Musk.
lots of technologies
Neuralink includes many overlapping technical efforts. The interface itself requires electrodes implanted in the brain. To connect these electrodes with the outside world, Neuralink uses a small piece of hardware implanted in the skull. This includes a wirelessly rechargeable battery and a low-power chip that collects data from the electrodes, performs some simple operations on it, and then transmits that data wirelessly.
Performing all this requires precision neurosurgery, and the company is developing a surgical robot to make this process safe and consistent.
At the other end of the process, neural signals need to be interpreted in near real time to understand what’s going on in a particular brain region. This requires computer systems that can handle everything from patient-to-patient variability to hour-to-hour differences in brain activity. Finally, in some cases, the device will need to send information back to the brain in a way that neurons there can interpret (immediately or after a learning process).
It’s… a lot of things. And the event has seen people talk about almost all of them. In most cases, the information was broadly similar to that shown the previous year. Various animals with implants have been shown to do anything from play. Ping pong manipulating cursors and typing using their implants – more examples than last year, but not radically different. Similarly, Musk talked a bit more about the processing capacity of the implant, which is now partially powered by an ARM processor. There are some indications of evolutionary progress, but no indication that it is close to a final design ready for submission to the Food and Drug Administration.
Perhaps the most significant difference compared to previous years is the level of detail involved in the surgical robot. This time, there was both an on-stage demo of the hardware and a lot of time spent discussing the details of the surgical procedure being developed for it. In the previous update, the development of the robot seemed to be delayed.
we’ve been here before
The event was said to be an overview of the company’s activities and the presentations seemed to cover all the key areas Neuralink was working on. But there are problems with this approach.
The first is that brain implants have been an active area of research for decades. While the details differ, a lot of what Neuralink shows has been done before. To some extent this is understandable. Neuralink develops its own electrodes, implant and processing system. As such, these systems need to demonstrate that they can perform like electrodes previously tested in animal experiments. But, at least so far, Neuralink has provided no indication that its systems are superior to those that have already been extensively tested or are on their way to getting there.
Meanwhile, some of its competitors have advanced in areas where Neuralink is trying to differentiate itself. For example, Blackrock Neurotech is now promoting fully implantable electronics that offer wireless charging and data transfer. And the company has already submitted hardware through a clinical trial and is applying for FDA approval. In fact, the company has several additional clinical trials ongoing.
The custom surgical robot appears to be unique to Neuralink (though surgical robots are commonly used for other purposes). However, one of the Neuralink staff cited a point of contention with the FDA, saying it was difficult to demonstrate the robot’s safety in a way that would satisfy regulators. And another of its competitors, Synchron, hopes to eliminate the need for major surgery by using blood vessels to insert implants deep into the brain. And these devices have already passed clinical trials.
Another problem with Neuralink’s progress update is that the company hasn’t made it clear that it’s ready to go to the FDA. Starting a clinical trial means that the company has finalized a hardware design (even if it’s working separately on next-gen hardware) and has chosen a specific neural defect it plans to treat. The update’s scattered image progress reports gave no indication that any of this was done.
None of this means that there will ultimately be no room for multiple technologies in the brain-computer implant space. Neuralink will likely eventually get to where some of these other companies are now, or find a niche where their hardware is particularly effective. However, so far, the company has not shared any information showing that it is close to either result;
Neuralink’s presentation is available online. Oddly enough, for an organization that claims to be a fan of free speech, the company has disabled comments on the video.