Want an Idea of How Far Computing Tech Has Come in Just 10 Years?

This is great and all, but I'd rather see things like AI develop than just more and more power. There has been disappointingly limited progress on that front.

I do not want to see Quark vibrate

AI will emerge from enough power.

Agreed. You have to have one to have the other. Sure, if you're a genuine Dr. Soong type super genius, maybe you can get some sort of AI running on an Apple IIe. But, it'd be the equivalent of a retarded cockroach.

And yet the best progress we've seen comes from brute force attacks, like Google's enormous neuronal networks.

Actually, it all depends upon how you define "AI." Back in Ought Two, I worked for a cellphone company that used an AI system to screen calls before they were routed to a human, a disturbing number of customers thought that the AI system was a real person. The real problem is we've talked about AI systems for decades, but never actually thought about what they were. When people talk about AI systems, they generally have a vague idea that it'd be something like a slightly less homicidal version of HAL, but do you really need your Roomba to be smart enough to play chess with you? No. You need it to be smart enough to vacuum your floor, which it is. If you make your Roomba too smart, then you risk turning it into Marvin the Paranoid Android ("Here I am, brain the size of a planet, and all I get to do is to roll around on your floor sucking up cat hair and dead skin. How I wish this house had some stairs so that I might throw myself down them."), and its a waste of computing power. You want a Roomba smart enough to vac your flat, and no smarter, otherwise it might start getting ideas that there's more to life than sucking dirt.

Google's self-driving cars are an AI system, and one that you wish had as much brain power as possible, because, not only does having a super-intelligent car mean you're more likely to be safe, but it can also keep you entertained during commutes and long trips.

I read an interview with an AI expert some years back and he claimed that there was actually tremendous progress in AI, but that people had stopped referring to the systems as having "AI" because when people heard the term "AI" they expected a non-murderous HAL, and would become frustrated if the system wasn't 100% perfect. But if you called your AI system something like "automated process control" people didn't expect to be able to talk to the system, and they were less likely to become incensed at any mistakes which cropped up.

I suspect we largely won't even talk about storage and memory. It will be so cheap and so readily available, that for all intents and purposes it will be infinite.

Yep, there's a strong likelihood of that. While there will certainly still be measurements for those things, it's quite plausible that they'll no longer be listed simply because they'll be largely irrelevant to anyone's purchasing decision.

To that, you may want to add 'connectivity.' I'm basing that on the difference between two things:

The 'AI' you get in console and computer gaming -- which falls into two categories that I'm familiar with, scripted (which isn't properly AI at all to my mind, but meh) and 'Reactive Agent' AI seen in "sandbox" style games like the Battlefield, GTA, Elder Scrolls, and Fallout series of games.

Contrast that rudimentary form of AI with things like this, where robots are actually learning from each other via networking.

So, yeah -- create machines with the requisite computational power, let them learn from each other, and the Singularity gets a whole lot closer to reality.

Along those lines: http://qz.com/189382/we-already-live-in-the-age-of-robots-we-just-dont-call-them-that/

You mean to say you had a pony in your ass?

I has a 3dfx voodoo card back in the day. I have a nvidia 660 ti in this machine.

Fuck me. NVIDIA just announced a 1 teraflop chip for phones.

A supercomputer in your pocket. A goddamned supercomputer in your motherfucking pocket!!!!![/i]