Today Elon Musk and his team at Neuralink made a presentation on their brain interface technology. Essentially they want to build a device that can be implanted in to the human brain that can restore bodily functions that have been lost either due to injury or a degenerative disease.
How important were the announcements made today? Why was the team so coy to talk about whether the current tech has been tested on an animal brain? How difficult will it be to solve the remaining problems before this type of tech has practical application? What are the known unknowns?
No idea how far past the state of the art this is, but I found the thread architecture to be very impressive. I’ve never heard of this being done before: using silicon lithography techniques to produce microstructured wiring. IIRC, each thread has a 6 micrometer diameter but contains 128 electrodes. Obviously there’s no way to bind together normal wires that way. Instead, they fabricate it using the lithography techniques of depositing thin layers of material, etching, and using photoresist to control where this takes place.
I think it’s pretty clear now that software will be the hard part. They’ll probably get to millions of electrodes in under a decade, but actually doing something with that data will prove hard.
I don’t know exactly what they are wanting to do, but what is happening is that they are asking the FDA to give them permission to begin human trials. What I expect to happen is that the FDA will say “No, we will not give you permission to drill holes in people’s heads and use a robot sewing machine to stab dozens of wires into their brains. Now get the fuck out of our office, you psychopathic twits.”
People get very worked up over animal testing, particularly primates. It’s obviously a necessary step here, but even so I’m sure they don’t want PETA protesting on their doorstep.
UC Davis, my alma mater and where Neuralink is doing their primate testing, very regularly gets groups of protesters. It comes with the territory, but that doesn’t mean they want to make too much noise about drilling holes in monkey skulls.
Ah well, that’s what they would do if you walked into their office and proposed a study in healthy volunteers. But nobody but a fool (or a psychopathic twit) would suggest that. If you can find a population of severely disabled, terminally ill patients whose lives might be improved or extended by this sort of thing, they’ll listen to you.
But mostly, these sort of supertech announcements remind me of British Satellite Broadcasting. Back in the 1980s, when Sky were rolling out their satellite dishes, BSB were competing with them. Their USP? *Satellite dish - Pah! - all you’ll need for our system is a shiny piece of paper the size of A4 (Letter in US, approx) - you won’t even have to have it on your roof, or pointed precisely at a satellite. Go up in the loft and stick it on the side of the chimney that most closely points to south - simple as that!
BSB is no more, taken over by Sky decades ago. Oh, and that flashy new technology? We’re still waiting for that.
I imagine if they’re really confident about the whole idea that they could start human trials in a country with a less stringent FDA equivalent. But, frankly, like many of Elon Musk’s ideas, it will probably not ever meet expectations.
There are already brain interface devices that provide rudimentary sense data in the deaf and blind. This will happen with or without Musk. His cult of personality makes everything about him clickbait.
Do the electrodes end up near synapses, axons or somewhere else? Are they activated by changes in ion concentration or something else? If electrical, could there be two way communication with the computer sending signals directly to the brain?
How to they prevent immune system rejection of the cables?
The probes end up where they end up–they don’t have that kind of fine placement yet. Mainly they just want to avoid hitting blood vessels.
It is entirely electrical, and two way. The current system doesn’t have as many stimulation channels as read channels, but they can be arbitrarily reconfigured–any electrode can act as a stimulation channel (just not all at once).
The new electrodes are part polymer, which I presume helps. I think the short answer is “material science”.
Scientific American has an article on Neuralink. Not a huge amount of detail, but it quotes a few experts in the field, who all seem pretty positive on the announcement.
The article mentions some existing products which I wasn’t aware of: the Neuropixel probe, or instance. It also appears to use lithography techniques for the multi-electrode tip. Although it has more channels per probe, it has a much larger cross-section (1700 µm[sup]2[/sup], as compared to Neuralink’s 30 µm[sup]2[/sup]). And Neuralink’s overall system has about 10x the number of electrodes.
So, seemingly still a good step in the state of the art. It looks like the robot installer will be crucial to expanding electrode counts further.