Jul 18, 2019

Elon Musk's plan to merge with AI

Illustration: Lazaro Gamio/Axios

For 5 years, Elon Musk has been warning about apocalyptic runaway AI, calling it more dangerous than nukes. To stave off his feared future, in 2016 he launched Neuralink, a company to create cyborgs with the express mission of getting ahead of superhuman intelligence.

What's happening: Now, Musk says he has charted the long path to merging man and machine. In an elaborate presentation Tuesday night, he said his company has installed brain–computer links in rats and monkeys and aims to put them inside human skulls next year.

The big picture: Around the world, top research labs are building brain–computer interfaces (BCIs), devices that can both read brain activity directly from neurons and write information straight into the brain.

  • At this early stage, BCIs are being used to treat conditions and injuries related to the brain or nervous system, including Parkinson's disease or paralysis, allowing people to control, and even feel, prosthetic limbs with their minds.
  • In the far future, researchers want to implant interfaces into healthy people. Among their ideas is to use the implants for communication, or a super-efficient connection to an electronic device.

But for Musk, medical uses are a stepping stone to an existential imperative.

  • Last year, he told Axios on HBO that Neuralink's ultimate goal is to "achieve a long-term symbiosis with artificial intelligence."
  • BCI, he says, is the best defense against an alarming future in which AI suddenly surpasses human intelligence and leaves our species behind — or totally imperiled.

Key quote: "This has a very good purpose, which is to cure important diseases — and ultimately to secure humanity's future as a civilization relative to AI," Musk said Tuesday night at the California Academy of Sciences in San Francisco.

Many AI researchers, who still struggle to get computers and robots to complete some basic tasks, disagree with Musk's techno-doomsaying. But Musk has called them "fools."

How it works: Neuralink's system consists of hundreds of electrodes implanted deep inside the brain, connected by tiny wires to a hub that communicates wirelessly with a device behind the wearer's ear. There's also a robotic "sewing machine" that plunges the electrodes into patients' brains.

  • A version of the device has been tested on rats, according to an unreviewed white paper Neuralink circulated after Tuesday's event.
  • And in response to a question on stage, Musk casually dropped that it's also been installed on a monkey, which used it to control a computer. The next step is human testing, which Neuralink hopes to get underway next year.

Reality check: Getting surgical implants into healthy humans is a long shot in the near future, says Kenneth Shepard, a BCI researcher at Columbia University.

  • For the Food and Drug Administration to approve an implant, the upsides have to far outweigh the risks. "And anything that requires surgical implantation comes with a lot of risk," says Shepard.
  • Much more likely to be approved are medical applications that address sensory or motor problems like blindness or paralysis — a significant benefit to outweigh the risks of poking stuff into brains.
  • And those are still out of reach of the best scientists, who are developing devices and software to reliably decode brain activity and send signals back into the mind.

Asked Tuesday how he plans to construct a viable business, Musk said simply that the economics of curing brain diseases and injuries "will easily pay for itself."

  • In the long run, Musk joked, "I think it's safe to say you could repay the loan with superhuman intelligence."

Go deeper: A device to translate brain activity into speech

Go deeper

Podcast: Elon Musk's plan to merge man and machine

Elon Musk on Tuesday revealed new details about Neuralink, his brain-computer interface company that first aims to help those with brain injuries and then protect humanity from AI. Dan and Axios' Kaveh Waddell discuss.

Go deeper: Elon Musk's Neuralink computer brain implant plan

Keep ReadingArrowJul 18, 2019

The tricky business of improving human brains

Illustration: Sarah Grillo/Axios

Connecting brains directly to machines has helped paralyzed people begin to speak and amputees feed themselves again — early steps toward the miraculous cures that have been the main focus of the neurotechnology field.

But a smaller group of researchers and startups — plus the Pentagon — is working toward an even longer-term goal fraught with scientific and moral hurdles. They plan to improve on healthy humans, in a bid to pick up where evolution left off.

Go deeperArrowJul 25, 2019

Looking to AI to understand how we learn

Illustration: Aïda Amer/Axios

Two parallel quests to understand learning — in machines and in our own heads — are converging in a small group of scientists who think that artificial intelligence may hold an answer to the deep-rooted mystery of how our brains learn.

Why it matters: If machines and animals do learn in similar ways — still an open question among researchers — figuring out how could simultaneously help neuroscientists unravel the mechanics of knowledge or addiction, and help computer scientists build much more capable AI.