View Single Post
Old 11-03-2015, 10:36 AM   #63
peter12
Franchise Player
 
peter12's Avatar
 
Join Date: Jul 2002
Exp:
Default

Quote:
Originally Posted by Phanuthier View Post
maybe the gate size has shrunk, but how about other materials? GaAs? how about using photons instead of electrons? Moore's law no longer isn't going to be the limit in optics.

why do you consider Moore's law to be a limit, when it comes to AI?

what software problems do you see as the road block (instead of speed bump)
As far as I have read (I am not an engineer), you max out at the atomic level, even with advanced materials, and electrons start bleeding all over the place. If you had a quantum computer, then yeah, things keep rolling forward. Only problem is that a quantum computer seems to be a long way off.

Software is still written by human beings. Computers are pretty dumb outside of the tasks written for them by humans. At this point, they seem pretty good simulating, and then executing responses based on the increasing ability to recall, and reorganize lots of information. This is why we all love algorithms so much.

I guess it remains to see if we have picked all of the low-hanging fruit. A lot of the new "profitable" software platforms are just hyped-up sharing platforms. They have always existed in one shape or form, but are now being revived under this not so new idea of the sharing economy, which is really just Silicon Valley VCs figuring out a new way to become rentiers.

As I keep saying, the implementation of technology has a lot of barriers - political and economic. When we start to understand these barriers, we see that even the idea that technology is inevitable is in some way a political way of viewing technology. Look at the graveyard of Google products. Look at something like Google Glass - a few years ago, it was supposed to revolutionize the way we experience reality. Humans rejected it, almost en-masse. Oculus Rift is the same thing, basically an iPad strapped to your face.

I am reeling way off-topic here. There is a lot to say, but to tie it back to autonomous cars, yeah, we have had the technology for like 20 years. Didn't Carnegie Mellon's robotic department drive a Pontiac across the state on autopilot in like 1993? Pretty much robot cars can do anything in a straight line. That is why they have had, for a long time, limited use in industrial settings. A robot truck can drive from point A to point B and back again day and night. But when we get into the real issue of widespread adoption of such technology, the questions become way more complicated.
peter12 is offline   Reply With Quote