I believe AI is a great tool that we should develop further. But we need to be smart about it. Power-usage is stupidly important, and researchers have been looking at spiking neural networks (SNNs) for a while as a lower-powered option which more closely mimics the way the human brain works. I’m not a ML-researcher, but to me that translates to “biglier smarts and smallier electrickimicity”. Maybe I’m wrong.
Now, from my drunk layman’s understanding, SNNs haven’t quite gotten to the point of giving outputs which are as useful as traditional artificial neural networks (ANNs). So there’s an obvious reason they haven’t become the standard. But does it really make sense for tech bros to invest billions building these datacenters across the globe for a model that is…likely to be obsolete, just because “AI” is the hype-word du jour? I don’t think so. Seems to me that it makes more sense to pour research money into SNNs. But what do I know, I’m just a country hyper-chicken.
I’ve wondered the same, but without even knowing about SNNs. I just wonder how long this kind of scale will be needed? Or will something akin to Moore’s Law or changes in software/algorithms make what is being built out (and the related power/water uses) irrelevant?
Apparently I got drunk and posted about SNNs. Awesome. Don’t take anything I said as, in any way, valid. I’m a moron and this is wildly outside of my domain. Apologies for any confusion I may have caused with my idiocy.
I believe AI is a great tool that we should develop further. But we need to be smart about it. Power-usage is stupidly important, and researchers have been looking at spiking neural networks (SNNs) for a while as a lower-powered option which more closely mimics the way the human brain works. I’m not a ML-researcher, but to me that translates to “biglier smarts and smallier electrickimicity”. Maybe I’m wrong.
Now, from my drunk layman’s understanding, SNNs haven’t quite gotten to the point of giving outputs which are as useful as traditional artificial neural networks (ANNs). So there’s an obvious reason they haven’t become the standard. But does it really make sense for tech bros to invest billions building these datacenters across the globe for a model that is…likely to be obsolete, just because “AI” is the hype-word du jour? I don’t think so. Seems to me that it makes more sense to pour research money into SNNs. But what do I know, I’m just a country hyper-chicken.
I’ve wondered the same, but without even knowing about SNNs. I just wonder how long this kind of scale will be needed? Or will something akin to Moore’s Law or changes in software/algorithms make what is being built out (and the related power/water uses) irrelevant?
Apparently I got drunk and posted about SNNs. Awesome. Don’t take anything I said as, in any way, valid. I’m a moron and this is wildly outside of my domain. Apologies for any confusion I may have caused with my idiocy.