You need Elon Musk’s new invention like you need a hole in your head.
Late Tuesday night, the Tesla and SpaceX founder unwrapped new details on work that would allow humans to transmit data straight from their brains to their computers. Musk’s project, called Neuralink, plans to start clinical trials in humans by the end of 2020.
The humans who choose to embed Neuralink wires in their brains next year should expect to have four small holes drilled into their head. Neuralink will thread wires into the brain and patch the skull holes with a computer chip, which can connect wirelessly to an iPhone app. The company touted its innovation around those threads as a big breakthrough toward a functional brain-computer interface.
In a white paper published on July 16, Neuralink claims to have “several advantages” over other brain-computer interfaces in development. The threads “are a better match for the material properties of brain tissue than commonly used silicon probes” and allow for greater flexibility about where those connections are inserted into the brain.
The research team writes that if the relevant “technological challenges” were overcome “it is plausible to imagine that a patient with a spinal cord injury could dexterously control a digital mouse and keyboard” with the system.
Musk billed the invention as the next step in human-computer connection.
“This is going to sound pretty weird, but ultimately, we will achieve symbiosis with artificial intelligence,” Musk said in the Tuesday press conference.
The company has already had some success in animal trials. In its press conference, Neuralink showed a picture of a rat it had wired up with the technology. The animal had a large USB device protruding from its skull, which Neuralink said will not happen when they install the devices in human test subjects. The company also unveiled a "sewing machine-like” robot that would perform Neuralink installation surgery.
Neuralink could help people with neurological damage communicate and control devices. But installing the technology will mean drilling holes in a wearer’s skull and lacing it with wires. And some ethicists and scientists question how Musk’s company will process the brain data and whether the devices might harm the brain.
Part of Musk’s motivation is his long-running fear of artificial intelligence, which he has described as humanity’s “biggest existential threat.” In a 2018 SXSW speech, he called the technology more dangerous than nuclear warheads and warned that super-powerful computers could soon overtake humans.
On Tuesday, he suggested in-brain computers might be a way to keep up with artificial intelligence.
“I’ve said a lot about AI over the years,” he said. “But I think even in a benign AI scenario, we will be left behind. Hopefully it is a benign scenario, but I think with a high-bandwidth brain-machine interface, we can hopefully go along for the ride.”
Apocalyptic scenarios aside, some experts have voiced concern about Neuralink. Neuroscientists have stressed caution about the devices, pointing out that Neuralink’s paper on its technology has not been peer reviewed.
Tim Harris, a medical researcher and developer of neural interfaces, told the Wall Street Journal that the devices should be able to last at least five years in the human brain: a litmus test Neuralink has yet to pass.
Musk has previously adopted an act-first, get-approval-later tactic for his technologies. In 2017, he infamously tweeted that he’d received verbal government approval to dig a tunnel between New York, Washington, D.C., Baltimore, and Philadelphia for his high-speed Hyperloop transportation. That approval has yet to materialize, and Musk’s tunnel projects are off to an underwhelming start. Musk’s previous tweets about Tesla stock prices resulted in a protracted legal battle with the SEC.
Musk’s company will seek FDA approval for the Neuralink technology, although the approval process will likely take a long time, he conceded Tuesday.
“Getting FDA approval… is quite difficult,” Musk said. “This will be a slow process where we gradually increase the issues that we solve until ultimately we can do a full brain-machine interface.”