4 ms·
This is hard, but perhaps even harder than you might think. I read an extremely long exploration of this concept a couple weeks ago. It's a layman explanation b
by lwansbrough 9y ago
This is hard, but perhaps even harder than you might think. I read an extremely long exploration of this concept a couple weeks ago. It's a layman explanation but it really does a great job of outlining the real complications of such a device. Here's the article: https://waitbutwhy.com/2017/04/neuralink.html https://waitbutwhy.com/2017/04/neuralink.html
One takeaway that gives you the sense of scale: there are at least 40,000 neurons in 1mm^3 of the cerebral cortex. Those 40,000 neurons create around 20,000,000 synapses. You have to create a sensor that can capture a reasonable portion of those synapses to understand what's going on in that tiny section of the cortex.
- criddell 9y agoModern CPUs have on the order of 10 billion transistors and we have a pretty good idea of how those work. In my (uneducated) opinion, it feels like current tech must be close to providing the tools and scale necessary to start figuring this stuff out, no?
- apl 9y agoWe didn't have to reverse-engineer the CPU. Also, CPUs are rationally designed; we have zero guarantees that an evolved brain follows any principles at all. Every brain region may be highly specialised to a single algorithm or task with zero mechanistic overlap. Figuring out deep neural nets is a closer analogy, and at the moment we have very little intuition for how to do even that. Keep in mind that the brain is vastly more complex than even the most advanced ANNs. On the technical side: The most advanced neural techniques allow for parallel measurement calcium signals in ~10,000 neurones. That's a long way off from complete observation!
- swiley 9y agoMost modern semiconductors are fairly flat (or are "3d" but it's the 90s autocad kind of "3d" where you have some number of layers.) Semiconductors are really fairly simple in comparison.