I’m living in my future. Didn’t expect this temporary regression but overall it’s progressing, if slower than I’d like.
Connectionism and parallel computing setbacks in the 90s slowed things down but it allowed for 4th gen computing to speed up at least but I still think that could’ve been done in tandem with 5th gen.
Still, not too bad for 2018.
What you’re using. 5th gen computing died in Japan, 1991.
5th gen is connectionist, parallel processing. AI based. Neural nets in hardware and software.
Instead, we developed the networks outside of the computers rather than within the boxes. Network of networks – internet – got the focus instead.
4th gen is basically single task / fast task switching like we have. Lots of caching, processes wait neatly in a queue.
Yup. A lot of shrinking though, which is fantastic. I also love the integration of network/graphics/sound/UI that you find in portable computers (android / tablets / iphones).
But I’m still awaiting significant architecture changes. It’s coming. Has to. Then again, with server processing with massive backup (cloud computing) perhaps fast enough speeds will render the need for AI computers to be moot.
But really, I’d love an intraconnected smart cube in front of me that needs no network to function as an AI. I’d probably want it to be built upon Practopoeisis concepts, although work in that area is a mostly “one man shop” – and he just got a job working for a blockchain service company.
This is a demo of Danko Nikolic ‘s concepts in action. I was sold on this a few years ago, and the relatively new demo proves to me that it’s probably the right approach.
But, has limited traction at the moment and older methods prevail.
Still, progress continues.