this post was submitted on 29 Jun 2024
1 points (100.0% liked)
SneerClub
989 readers
1 users here now
Hurling ordure at the TREACLES, especially those closely related to LessWrong.
AI-Industrial-Complex grift is fine as long as it sufficiently relates to the AI doom from the TREACLES. (Though TechTakes may be more suitable.)
This is sneer club, not debate club. Unless it's amusing debate.
[Especially don't debate the race scientists, if any sneak in - we ban and delete them as unsuitable for the server.]
founded 1 year ago
MODERATORS
you are viewing a single comment's thread
view the rest of the comments
view the rest of the comments
Why did you think it was working well with no problems?
Literally the first human-cpu interface?
You hecka optimistic.
I mean, it's pretty crazy how well the design did work considering it's the first of its kind.
The latest thing I saw, a bunch of the wires are becoming detached from the very first prototype, which of course is being worked into the subsequent models.
Pretty sure keyboards are older
@Varyk @pikesley
Not the first, actually a late entrant.
I worked in a lab using implanted brain-computer interfaces 14 years ago.
Other labs using the same system had monkeys controlling robot arms, and a human controlling a computer.
I seriously don’t know where these fuckers have been that they’re supposedly excited about implanted BCIs now that musk’s doing a monstrously shitty one, but somehow they managed not to ever read about any of the previous research into this that had the same outcome as neuralink (basic, inaccurate computer mouse control) with the same major caveats (the electrodes become unusable in short order due to scarring and can’t be repaired), except that the neuralink version is unnecessarily risky* cause startups gotta go fast
[*] and the risk here is that something truly fucking awful will happen to the patient, because it’s the fucking human brain and they’re treating it like a submarine made of secondhand carbon fiber, including the ignored track record of failed lab tests before disaster struck
I still don't think cortical electrodes should even be described as a direct interface. A direct interface would speak action potential and connect to your spinal cord, like Ghost in the Shell or The Matrix. I do not see a lot of people lining up to test what would probably entail having your brainstem dissected!
@self
A lab I worked in (as an IT guy) used them for data collection, studying visual attention in monkeys.
Not a happy place for the monkeys although I'm confident the scientists did their best to not make it any worse than it had to be.
I wish I had the confidence of a techbro who thinks any of the BCI tech in neuralink is new and isn’t just a set of techniques that have existed for decades and have a shitty track record
the only thing neuralink seems to add is wireless control, which doesn’t work, partially due to impossible bandwidth and compression requirements, but mostly because it’s a project driven by Musk’s whims
It still amazes me they publicly posted a request for help with these compression req which are physically impossible to achieve. Nobody with a CS degree is anybody near the leadership of neuralink. In other words, you are downplaying how impossible the requirements were.
What is the original requirement? I've never seen this and I feel a mighty sneer that I'm missing. A Fear of Sneering Out if you will.
Some 100 times+ compression, very fast and also at very low power (because it was inside the brain), and lossless.
And not stated but because it was inside the brain, I assume possible memory usage also wasn't that high etc. Sorry have no direct link to the specs.
@Soyweiser @V0ldek
The multi-electrode systems at the lab I worked in 2009 used a fiber connection to the host PC and generated terabytes of data, for just 128 or 92 electrodes (I forget) at not-all-that-many samples per second.
Sorry what is the point here? Or do you just want to share a cool story about data transfer? (I'm not trying to be snarky, I feel like im missing something here).
@Soyweiser
Just an example of the data involved in a brain-computer interface.
look, with the 200x compression,
ye iirc we can remember like 7 things so that might be hard /s
okay but if they hadn't done that we wouldn't have gotten this work of art
holy fuck I read the comments
all the musksucking sub-linkedin reply guys chittering at each other about how this is clearly the kind of breakthrough musk was hoping for when he devised this genius-level challenge. you know, basic lossy audio compression applied by an asshole pretending to be an audio engineer (or maybe this fuckhead is the reason the FLACs I pay for are sometimes just converted from MP3s?)
e: I got to the post where he supposedly makes a point about ADC and container bit depths. this fucking idiot is almost certainly an audiophile cosplaying, and the only audio engineering he’s done is set his amp on a block of wood
oh absolutely! but only out of a sense of shame for being in a career field where a medical device company posting that horseshit compression challenge didn’t immediately prompt a strong backlash and repercussions for neuralink’s ability to attract and retain talent, in lieu of a functioning regulator maybe possibly shutting them down before they can fucking mutilate someone else with this brainfart of an invention
I feel bad for anyone who gets that e-waste implanted into their head and ends up with an implant that absolutely cannot do the things it’s marketed to do, barely does ordinary 90s brain implant shit, stops working very quickly (to the apparent surprise of the people in charge) and will most likely cause injury and severe discomfort to the patients saddled with it
I wish my field had ethics. I’d sleep better if we did.
you do not have much medical knowledge, do you?
@Varyk oh wait so it *doesn't* work? But it's still a good idea to attach it to your brain?