Discussion in 'Science & Technology' started by Damn_Smooth, Aug 18, 2011.
It's only a matter of time now.
I, for one, welcome our new microprocessor overlords.
All hail master Bender.
Interesting, but does this remind anyone else of a PGA (programmable gate array)?
It sounds similar, but I'm nowhere near educated enough on the subject to comment on differences.
Doesn't the programmable gate array have to be manually programmed, though?
Yes. IIRC, it's wired so that you burn out the connections you don't want. I just meant in terms of the concept.
Although even if you could reverse the process on the fly, this sounds different in another way too. According to the article, it doesn't just enable or disable certain connections, but gives each one a weight. So it might not use a particular connection very often, but it doesn't ditch it completely either. That's how I read it anyway. Ofc they can't give the media too much detail at this point.
wait could this mean upgradable cpu's
in the next few decades maybe if the research has good findings
Ewwww that's disgusting. Now machines will play with themselves. Gross.
The problem with everything that can program itself is that it may become useless to the person who buys it. I don't think learning on the electrical/conductor level is wise. I think software being allowed to rewrite portions of its code to better handle a task or perform an entirely new task would be preferable. Yes, it's slower but it is also more predictable and controllable.
it will learn about us and turn us all into cyborgs. "resistance is futile".
clearly, its possible to set limits on what they can learn/change. there would also have to be 'resets' so the device cant learn itself into being slower for whatever reasons.
Only higher function code can establish limits that are necessary. Something as low-level as a processor is only effective at protecting itself from itself (e.g. NX).
By useless, I didn't mean slower. I meant a chip that reprograms itself found in a robotic arm, instead of performing the task given, would end up breaking everything because the decisions it decided to make on its own were not in line with those that bought it/programmed it initially. Similar examples can be given for every field.
Computers role in this world today is to obedient servants and, what I described, allows them to do that better with few risks. What IBM is trying to achieve is more like an experiment in building a sentient computer.
but they have rights .....
If we assume for a second that humans are walking bio-chemical machines and if they get their true potential "locked and limited" by some creator then it's the same. It's unfair.
They already exist. Thank you, Intel. </trollpost>
Yeah ... until they override the "Three Laws of Robotics" programmed into them and start viewing us as expendable meatbags.
Time to buy more large caliber weapons and armor piercing ammo.
If the cyborgs decide to take over at the same time the zombie apocalypse starts, we're going to be in a world of hurt.
Zombies - Bad, but slow and easy to kill.
Cyborgs - Bad, fast and tough, but possibility to reprogram for our team.
Zombie Cyborgs -
Zombie Cyborgs! Kreij, you're a genius! We need a movie about those.
Jame Cameron is gonna sue them for infringing his patents.
Actually. Wait. Do our new Cyborg overlords look like Summer Glau ? If so, I'm all for it
A really good FPS movie tie-in game would be cool. (key words being really good).
Here's a question ...
Since both zombies and cyborgs don't kill their own kind, if a Cyborg killed a Zombie, who would the Zombie Cyborg side with?
Isn't this a bit like like the alien vs predator and at the end they made a hybrid and so they both killed the hybrid.
Okay ... now we have highly armed Zombie Cyborgs too.
Too bad the article didn't have more detail. It would be interesting to read an overview of how the "learning" algorithms function and make determinations based on experience.
Separate names with a comma.