Hot damn, this got a lot of points, but no discussion.
The basic TL;DR is that they figured out a thing that works really, really well as a neural implant material (a mesh). They spent a lot of time figuring out the ideal details for this mesh, and a bunch more time looking at how it interacted with the brain.
Here's some TL;DR from the bulk of the paper:
Normal implants have the issue that the don't move the same way as brain tissue, so anything that jostles the brain causes the implant to carve paths through the brain. Since this is a mesh, it moves with the brain.
Injecting a "normal" mesh, you just get a sort of tube of stuff where you injected it. Because of the shape of this mesh, it tries to uncoil; but does so slowly enough that it doesn't damage the brain tissue. One of the major points of the study is the specifics of this mesh (which is basically just an angle).
Because it's a mesh, neurons actually really like growing on it, so they form a bunch of handy connections to it.
One fun thing you can then do is grow brain tissue in the petri dish with the mesh, and you'll get something with real nice connections.
As an aside, this is actually the first paper that I went: "Fuck it, I'm going to read Science"; that it was actually so unexpectedly accessible has started a trend of actually reading the science behind things. I recommend it!
(Disclosure: I are software dev and do not know medical anything but can read good, so I just do a lot of that)
Neural implants ALREADY re-wire muscles; more, they can actually stimulate nerve regrowth (you trying to use those paths actually help them recover; Kill Bill style). I don't see this improving that effect, except that it's a better form of implant (less downsides)...
...however, I wouldn't be surprised if it was better enough than existing implants that you can do something new; namely, make artificial nerves to replace damaged ones... and I wouldn't be further surprised if the effects of the mesh act as a scaffold to promote cell regrowth, like the 3D printed organs.
LTP is a wonderful thing. Your brain grew that network before and now it is fading away. The only thing stopping you from replicating it is the missing medium and if we provide that before you lose the network, it may be able to.
It's been a while since my various neuroscience lectures/courses, but as far as I remember LTP only refers to the process of strengthening existing synaptic connections, not the forming of new ones. The process of forming new connections is called synaptogenesis.
That's just nit-picking about terms though, it is clear enough what you mean - except that growing the exact same network still doesn't give you the exact same synaptic strengths of each connection. And I don't know what you mean by replicating the existing network - do you mean a technique to recreate the exact same neural network? But even if you manage to do that, where are the boundaries, how do you know which connections matter and where do you stop? I probably don't understand your plan though.
Yes, we already have the tech for interfacing nerves. This way nerves can be reached with reasonable precision without surgery, likely without a visit to a hospital.
Enough that it is not considered a viable approach for general purpose BCIs in adult subjects. You can get away with injecting one to study or affect a small localized region, but that's about it [source: chat with a Lieber lab collaborator].
Unless of course you get Cultured and put it in place at the embryonic stage, so the brain grows into it.
Having a computer and radio in your brain doesn't mean you have to expose access to anything via the radio link. Even if you want to, there has to be software in between the connection to the meat and any APIs you provide for controlling your body, mapping intentions to raw signals for the neurons. This would have some form of access-control and safety features.
A general AI, especially a "malevolent" one, is such a mind-blowingly powerful entity that these questions are rather pointless. A self-improving GAI could rapidly take over the planet and have a myriad ways of "controlling humans".
I'd be far more concerned with other humans messing with it. And before that, just bugs and other damage.
The basic TL;DR is that they figured out a thing that works really, really well as a neural implant material (a mesh). They spent a lot of time figuring out the ideal details for this mesh, and a bunch more time looking at how it interacted with the brain.
Here's some TL;DR from the bulk of the paper:
Normal implants have the issue that the don't move the same way as brain tissue, so anything that jostles the brain causes the implant to carve paths through the brain. Since this is a mesh, it moves with the brain.
Injecting a "normal" mesh, you just get a sort of tube of stuff where you injected it. Because of the shape of this mesh, it tries to uncoil; but does so slowly enough that it doesn't damage the brain tissue. One of the major points of the study is the specifics of this mesh (which is basically just an angle).
Because it's a mesh, neurons actually really like growing on it, so they form a bunch of handy connections to it.
One fun thing you can then do is grow brain tissue in the petri dish with the mesh, and you'll get something with real nice connections.
As an aside, this is actually the first paper that I went: "Fuck it, I'm going to read Science"; that it was actually so unexpectedly accessible has started a trend of actually reading the science behind things. I recommend it!