The brain-computer interface race is on, with AI speeding up developments

Each company in the BCI field has slightly different ideas and methods for reading and processing human brain signals.

Synchron, which is just finishing a followup of its 10th patient, who has been living with its device for 12 months, and preparing for a larger clinical trial, reads brain signals by inserting a stent into the jugular vein that can read brain signals.

The benefit of the stent is that it can be inserted with a minimally invasive procedure. The downside is that, because the stent is not inserted directly into the brain, it doesn’t receive as many brain signals, giving patients less control over computers.

Precision Neuroscience, which is just beginning the FDA process, has a different method. It plans to make a tiny slit in the skull and slide an extremely thin set of electrodes in. It has also been testing its device in a novel way, recruiting brain surgery patients to temporarily test the products during their procedures. Precision then runs a series of tests on the patients, gathering invaluable data that can also help doctors performing the surgery.

Blackrock Neurotech, founded in 2008, has been testing implants for longer than any of its competitors and may be the first to achieve an FDA-approved device. The company’s devices are placed directly on the brain, giving patients high fidelity control. However, the devices have often lasted only a few years, in part due to scar tissue building up around the sensor, which degrades signals.

Neuralink’s device works differently. The sensors that collect brain data are fibers thinner than a human hair. The company developed a surgical robot to insert the sensors into the motor cortex. The device can then wirelessly beam the signal to a computer, whereas devices like Blackrock’s use a cable.

There are other startups, some toiling away in stealth mode. Neuralink’s co-founder and first president, Max Hodak, left in 2021 to launch his own computer-brain interface company called Science.

In my end-of-the-year tech predictions, I highlighted computer-brain interfaces, projecting they would make new strides in 2024. It has happened faster than I thought.

In November, 2022, I exchanged iMessages with an ALS patient named Rodney Gorham, who had been implanted with Synchron’s Stentrode and was using his brain to send the texts. It was an almost emotional experience. But it was also very much a beta product. It took Gorham several minutes to send a one-word message.

Synchron’s product roadmap includes potentially adding more veins to capture more parts of the brain, which would improve functionality.

Neuralink’s strategy is more of an all-or-nothing approach. Rather than create a minimally viable product and then improve it over time, its first version could be a game changer — if it keeps working and passes the FDA approval process.

We don’t know yet how the brain will respond to the device a year or two from now, and whether scar tissue that may develop will interfere with the signal.

If Neuralink and other BCI companies are successful, we will see these sensors expand to capture more of the brain’s neurons. Some people think or even hope that advancements will allow a melding of humans with machines.

For now, it will have the most impact on patients with debilitating conditions that prevent them from moving or speaking. Last year, we saw a proof of concept for what restored speech might look like in a pair of studies at Stanford and the University of California, San Francisco. Using artificial intelligence and BCIs, researchers enabled ALS patients to talk in a way that sounded natural. Those implants, connected to massive computers, were only temporarily implanted, but it may not be long before something like that is possible in a small, wireless device.

Another near term possibility is that patients will be able to control prosthetic limbs with their brains. This is also something that’s been demonstrated as a proof of concept. With devices like Neuralink’s BCI, we may soon see people walking around on artificial limbs as if they were real.

Elon Musk caused the price of sending a rocket into low earth orbit to drop 100-fold. It wasn’t because of a big scientific breakthrough. It came after blowing up a lot of rockets in a trial-and-error approach to innovation that resembled software development cycles. The move-fast-and-break-things method ended up working well with rockets, but it rubs scientists the wrong way when applied to experiments conducted on animals and humans.

Arthur Caplan and Jonathan D. Moreno argue in the Hastings Bioethics Forum that Musk’s decision to simply tweet about Neuralink’s human trials is an example of a lack of transparency: “Opening up the brain of a living human being to insert a device, particularly someone with serious medical problems, deserves more than a two-sentence report on what is, in effect, a proprietary social media platform not distinguished for its reliability where facts are concerned. This human research subject, their family, all potential future research subjects and patients, the medical community, and all of us deserve more.”