ARGH! The irony that after typing for an hour about computer technology the computer crashed when I was (actually addressing the software side
of things) in the DARPA website and clicked on a video that went straight to WMP instead of an IE window.
So my response will be a bastardized outline compared to the original response I was nearly finished with. But first, all note that the AI (AGI)
issue doesnt dismiss my main premise in this thread, in arguing that AGI is decades away.
Bastardized summary as follows:
Some newer semi-revolutionary computing hardware technologies, many of which are beyond the theoretical stage:
Quantum Computers
Photonic Computers
3D Silicon Processors
Component-level Optical Connections
Live Neuron Biological processing and hybridizations (
an
example
)
Software is the other side, and many prominent AI pioneer types believe that the right software may be easier than most would assume, and that the
hardware side may not require as much with the right software.
To better understand the possibilities in this world look deep into the
Law of
Accelerating Returns (LoAR).
Beyond exponentially increasing gains in computing power and related technologies (along with virtually every other facet of other human technological
progress), the key to AGI are
systems that can
learn. Some successes in terms of software alone learning (with basically COTS hardware
systems) consider DARPA's successful
PAL, and the de facto government operation Google
(read
here &
here).
When considering any of the more advanced hardware technologies listed above, there's no telling what the military and related higher institutions of
the technocracy, but we can gauge how to perceive the AGI subject by acknowledging some choice programs openly admitted by the military. Check out
DARPA's "global brain"
SyNAPSE
(being conducted by IBM), along with the rest of the projects listed out on
DARPA's IPTO website.
Don't factor out Google in your concepts. This has been an oversimplified version of my original response, sorry. The real thing to watch is more a
system showing the ability to learn than number crunching.