Billionaire technologist Elon Musk announced this week that his company Neuralink has implanted its brain-computer interface into a human for the first time. The recipient was “recovering well,” Musk wrote on his social media platform X (formerly Twitter) on Monday evening, adding that initial results showed “promising neuron spike detection”—a reference to brain cells’ electrical activity.
Each wireless Neuralink device contains a chip and electrode arrays of more than 1,000 superthin, flexible conductors that a surgical robot threads into the cerebral cortex. There the electrodes are designed to register thoughts related to motion. In Musk’s vision, an app will eventually translate these signals to move a cursor or produce text—in short, it will enable computer control by thinking. “Imagine if Stephen Hawking could communicate faster than a speed typist or auctioneer. That is the goal,” Musk wrote of the first Neuralink product, which he said is named Telepathy.
The U.S. Food and Drug Administration had approved human clinical trials for Neuralink in May 2023. And last September the company announced it was opening enrollment in its first study to people with quadriplegia.
On supporting science journalism
If you’re enjoying this article, consider supporting our award-winning journalism by subscribing. By purchasing a subscription you are helping to ensure the future of impactful stories about the discoveries and ideas shaping our world today.
Monday’s announcement did not take neuroscientists by surprise. Musk, the world’s richest man, “said he was going to do it,” says John Donoghue, an expert in brain-computer interfaces at Brown University. “He had done the preliminary work, built on the shoulders of others, including what we did starting in the early 2000s.”
Neuralink’s original ambitions, which Musk outlined when he founded the company in 2016, included meshing human brains with artificial intelligence. Its more immediate aims seem in line with the neural keyboards and other devices that people with paralysis already use to operate computers. The methods and speed with which Neuralink pursued those goals, however, have resulted in federal investigations into dead study animals and the transportation of hazardous material.
Musk has a habit of suggesting big things but providing little detail, notes Ryan Merkley, director of research advocacy at the Physicians Committee for Responsible Medicine. “This is maybe the biggest example of that” because there’s no information available about the person who received the implant or their medical condition, Merkley points out. “Depending on the patient’s disease or disorder, success can look very different.”
Scientific American spoke with Donoghue to understand what the latest step signifies for Neuralink—and whether the company might ever achieve Musk’s more extreme goals.
[An edited transcript of the interview follows.]
In describing the first results, Musk used the phrase “promising neuron spike detection.” What might that mean?
I have to preface this by saying that, as a scientist, you can’t comment on anything until you have a paper in hand. But what that generally means is: there are action potentials [the electrical impulses that nerve cells create], and there’s a probe in the brain, and it’s picking up signals that are there.
But I have to remain open to what that actually means. I have to see the data. And commercial entities do this piecemeal-feeding things. The danger with that, of course, is: tomorrow it may stop working.
Are you concerned by how the recent information on this trial was drip fed on a social media platform or anything else about how the research has been communicated to scientists or the public?
I don’t particularly like it, but I don’t want to overstate that…. It’s legitimate to say you implanted someone. I don’t think they want to say much about what it means or what it does.
From where you’re sitting, what can Neuralink bring to the table beyond the research that you and others have done—for instance, your work on enabling a participant to move a computer cursor on a screen by thought? Does Neuralink have the opportunity to do something different?
They’re a commercial entity, right? They want a product. When I started, I had a company called Cyberkinetics, which was too early. [Cyberkinetics was acquired in 2008 by what is now Blackrock Neurotech.] One of the differences is something on the order of $100 million. In those days, we lacked all the money and the knowledge. Now we have the knowledge. [Musk has] put in the money.
What he’s done is taken in investments, taken all that knowledge that the field has created. Certainly [that means research done by] not just us but [also] lots of [other] people, basic scientists who have figured out how the motor cortex works, including us but definitely not exclusively us, [as well as neuroscientists] Apostolos Georgopoulos, Andy Schwartz—all those people built a foundation. He could take all that knowledge and put the money in to create the kind of product that could potentially be commercial, though that’s yet to be determined.
I’m really happy to see that there will be (likely, we’ll see what happens with this particular version) a commercial product that will someday reach people and help them.
In terms of Neuralink’s approach, what’s novel about having a chip talk wirelessly to a device versus other proof-of-principle lab demonstrations? Previous implants have captured signals from individual neurons, but this may require cables to transmit the large amount of data involved.
That’s an important step to get everything inside. The problem has been that it’s a lot of information. And they came up with a solution that was more practical than I was thinking. I was hoping to get [a] full-bandwidth, high rate of information. They’re using Bluetooth, which is a much reduced version…. But they get enough information to be able to control things.
With the information they take out, they’re not able to pull apart every single neuron out of the whole mess. They mix things together a little bit, because the bandwidth doesn’t allow them to sort out everything. But it works.
After this announcement, are we any closer to Musk’s loftier goals? He has talked about restoring eyesight to people who are blind and mobility to those who are disabled and even about this wild AI-merging scenario.
Somebody asked [bioengineer] Ed Maynard, who was my graduate student in 1999, exactly that kind of question. He said, “We have modest goals: we want to make blind people see, paralyzed people move and deaf people hear again.” So that’s an old line that’s been kicking around for 25 years. And of course, everything we do toward improving the ability for people to have these things makes us a step closer to helping paralyzed people communicate and move again in a practical way, [with a] commercial device.
The thing is, you’ve got to be careful. The whole nature of restoring sensory inputs, like vision, involves electrical stimulation in the brain. It’s a whole different ball game. It’s not recording from single cells—that’s one thing–it’s stimulation.
And as far as I know, there’s not one scintilla of evidence of using that device to create sensory systems in any way. It’s a whole other project. You could say, “You’ve got a car. Do you think [a boat] will work?” Who knows? Yeah, it could, but it’s not a car.
They would need to outline what they think are the steps. Can you put electrodes in the brain? Yes. Can they stay in there? Yes. Can you stimulate them? I don’t know—they’re supposed to be able to. Can you stimulate them in the right way? Vision has been very complicated in terms of putting electrodes into the brain and trying to restore a meaningful image.
Are there limits to what a brain-computer interface can offer? Is it beyond the motivation that Musk has stated? Do we still have that dream of what folks were talking about 25 years ago?
The nature of science is: you never know what’s around the corner. I’m sure you saw Oppenheimer.
There were two potential outcomes of that [first detonation of an atomic bomb]: you’d get a really big boom, or you’d destroy Earth if it didn’t stop. They didn’t know. We don’t know! There are lots of things that have happened that I wouldn’t have ever predicted. The use of communication is much better than I thought it would ever be.
The engineering stuff is: Will we get a low-power amplifier that’s much smaller, that will have more bandwidth? Almost certainly, but you’re not talking to an expert in that area. It looks to me like we’re getting better and better with the capabilities of electronics. The engineers can tell us, “These are not insurmountable problems.” We have the technology already. And then there are other things that are going to be hard to do. They’re scientific questions [for which] I have no idea what the answer will be until we do the experiments.