71–80 of 80 posts
Earlier quoted context omitted.
Indeed. He also launches supposedly self-driving passenger cars into concrete barriers. It’s the latter that’s causing some concern I’d think.
Is it more or less concerning than when humans do the same? Autonomous driving is still a relatively young beast. I don't have specific numbers infront of me, but I imagine mile for mile, autonomous vehicles are still safer than humans.
one feature i dont understand about neuralink is the promised transmission of thoughts. lets say i have a device that can encode activity of every neuron in my brain simultaneously. i think some variation of the thought "elephant," and the neuralink creates some digital representation of these signals. then i send that digital encoding to your brain's neuralink. how on earth are you going to think of ANY elephant, le…
I'll start listening when they have a working definition of what a "thought" is or what "translating" it into a "universal language" even means.
I am not an expert, but what I thought was remarkable is the 12000 channels Musk talked about. New cochlear implants have 120 channels - one channel is one neuron (AFAIK) - more neurons is better results. I would love some expert take on this, but my first reaction was „12000 sounds like a lot - they all work? That’s cool“. Also check out this wait but why article which I think explains the resolution pretty well: ht…
I’m very much not an expert. I think the state of the art for measuring brain activity is the functional MRI. fMRIs have limits on their spatial, temporal and ‘activity’ resolution. I think they also measure ‘only’ blood flow and not electrical activity. My guess is that if Neuralink does nothing more than provide a more sophisticated tool for investigating the brain than the current state of the art then it will hav…
I've seen single-unit recordings done with electrodes and simultaneous Ca2+ imaging with a 2-photon microscope. You trade real-time speed and single unit specificity for a loss of knowledge about the population.
MEG can give you real-time summed behavior, but the inverse problem is indeterminate so you're less sure of _where_ that activity is coming from.
"what's the good tool to measure brain activity" is very often a function of "what do i want to find out about the brain?"
Earlier quoted context omitted.
> He also launches supposedly self-driving passenger cars into concrete barriers. Planes used to fall a lot in the beginning, computer hardware was once finicky and unreliable, even rockets used to blow up before they got it right. Takes time, and sometimes it takes early adopters trust to get to safety.
How many public deaths is an acceptable number? What about those dead who, rather than being "early adopters", just happened to be in the way of one - is there a line in the sand where we say "OK, we've sacrificed enough humans now, let's try something else"?
Earlier quoted context omitted.
Is it more or less concerning than when humans do the same? Autonomous driving is still a relatively young beast. I don't have specific numbers infront of me, but I imagine mile for mile, autonomous vehicles are still safer than humans.
I also don't trust random humans to drill holes in my head.
Prof Andrew Jackson, Professor of Neural Interfaces, Newcastle University, said: > “Neuralink are progressing through the steps that have been taken with previous neural interface technology. They have moved from rodents to a large mammal (pigs), and are apparently seeking FDA approval for a human trial. I had some sympathy with Elon, having also had experience of animal experiments that don’t work perfectly the day…
Well, and this is exactly what needs to be done to further this area as it is the part that's not completely addressed by academia. Any kind of academic research obviously focusses on generating one-time results for publication.
Prof Andrew Jackson, Professor of Neural Interfaces, Newcastle University, said: > “Neuralink are progressing through the steps that have been taken with previous neural interface technology. They have moved from rodents to a large mammal (pigs), and are apparently seeking FDA approval for a human trial. I had some sympathy with Elon, having also had experience of animal experiments that don’t work perfectly the day…
> “So in summary I would say this is solid engineering but mediocre neuroscience. Finally, I think it is unfortunate that they are presenting their work in this way, rather than publishing peer-reviewed papers that would allow their claims to be scrutinised, but I guess this is something that we will have to get used to as neural interfaces move from the academic to the commercial sector.” Well, and this is exactly w…
Earlier quoted context omitted.
On the other hand, SpaceX runs rather professionally. It also has to, of course. My point is, they CAN deliver. Probably with Tesla they're just trying to get by with the least acceptable effort.
SpaceX is somewhat less impressive as a company than you might imagine. They don't really have that much competition apart from Soyuz at the moment. There is huge support in US to have a alternative to Soyuz. Also NASA is prisoner of government corruption to the point where NASA operations are crippled in their capabilities - ie you get funding only if you keep unnecessary bureaucracy running for sake of some governo…
Earlier quoted context omitted.
> how on earth are you going to think of ANY elephant, let alone my elephant? Both brains sensors are calibrated on the same images, then a common representation can be found as intermediary. It has been done with neural nets translating into a common intermediary language, thus needing N encoders/decoders instead of (N-1)^2 separate networks. In the easiest case it can be a simple linear transform to map one vector…
wouldnt you need a distinct transformation for every thought?
you'd need to train a bidirectional adapter for every person instead of (N-1)^2 adapters for each pair of people
Earlier quoted context omitted.
wouldnt you need a distinct transformation for every thought?
the transformation of brain signals to the common space can be a neural net - and since neural nets are non-linear they can adapt to the inputs you'd need to train a bidirectional adapter for every person instead of (N-1)^2 adapters for each pair of people