Advantages Of Silver Nanowires As NNs

[towforce]

Good article by a couple of academics at Sydney University who have made NNs using silver nanowires - link.

I think it's unlikely that anyone is ever going to make a competitive chess computer this way, but the article is interesting because their NN behaves more like a natural NN than the ones being used to make breakthroughs in AI right now do, which, according to the article, confers some advantages:

* current AI requires huge amounts of energy and other resources for training

* current AI's ability to adapt and function in dynamic, hard-to-predict and noisy environments is poor in comparison to humans

* current AI lacks human-like memory capabilities

Do we need to use silver nanowires to get these advantages though, or could it be done in software simulation? Maybe the the silver nanowires would be faster due to their "parallel processing": the article doesn't cover this. They would have to be good for investors to pay to scale them up to the size needed for competitive AI though.