Skynet just came a bit closer

biostud

Lifer
Feb 27, 2003
18,193
4,673
136
IBM researchers make a chip full of artificial neurons

The new processor, which the team is calling TrueNorth, takes a radically different approach. Its 5.4 billion transistors include over 4,000 individual cores, each of which contain a collection of circuitry that behaves like a set of neurons. Each core has over 100,000 bits of memory, which store things like the neuron's state, the addresses of the neurons it receives signals from, and the addresses of the neurons it sends signals to.

They found that TrueNorth cut energy use by 176,000-fold compared to a traditional processor and by a factor of over 700 compared to specialized hardware designed to host neural networks.

"by tiling multiple TrueNorth chips, creating systems with hundreds of thousands of cores, hundreds of millions of neurons, and hundreds of billion of synapses."

http://arstechnica.com/science/2014/08/ibm-researchers-make-a-chip-full-of-artificial-neurons/
 

Jaepheth

Platinum Member
Apr 29, 2006
2,572
25
91
That's all well and good, but when can I plug it into my brain?

I want my neural pattern to live forever dammit!

Or at least be able to upgrade my memory or install a math coprocessor.
 
Mar 16, 2005
13,864
108
106
hr_Transcendence_4.jpg
 

JulesMaximus

No Lifer
Jul 3, 2003
74,456
854
126
Why do we always assume that a self-aware artificial intelligence would want to kill all humans? Maybe we would develop a symbiotic relationship with them.
 

MongGrel

Lifer
Dec 3, 2013
38,751
3,068
121
Yeah, saw that earlier.

Give em a few years it will get really usable.

Low power too.
 

crashtech

Lifer
Jan 4, 2013
10,521
2,111
146
Why do we always assume that a self-aware artificial intelligence would want to kill all humans? Maybe we would develop a symbiotic relationship with them.
Hard to know for sure, isn't it? We may see it in our lifetimes; I'm sure it will be an interesting experience. Hopefully such an intelligence will perceive some self-interest in preserving mankind, hopefully it will not then decide there are far too many of us for its purposes. Who knows.
 
Feb 4, 2009
34,494
15,729
136
I wonder about day to day stuff like:
Could someone face a murder charge for deleting it
Will AI's be able to vote, what about identical copies?
Would AI's make better politicians? No bribery
What about crime & punishment?
 

Jaepheth

Platinum Member
Apr 29, 2006
2,572
25
91
Why do we always assume that a self-aware artificial intelligence would want to kill all humans? Maybe we would develop a symbiotic relationship with them.

Projection.

I'd kill all humans (or a vast majority of them) if I could, and I'm one of them.
 

Crono

Lifer
Aug 8, 2001
23,720
1,501
136
Sounds perfect for future smartphones. Watch Dogs, anyone?

"To give some sense of its actual performance, the researchers got TrueNorth to do object recognition on a 30 frames-per-second, 400×240 video feed in real time. Picking out pedestrians, bicyclists, cars, and trucks caused the processor to gobble up a grand total of 63 milliwatts."
 

AyashiKaibutsu

Diamond Member
Jan 24, 2004
9,306
3
81
Sounds perfect for future smartphones. Watch Dogs, anyone?

"To give some sense of its actual performance, the researchers got TrueNorth to do object recognition on a 30 frames-per-second, 400×240 video feed in real time. Picking out pedestrians, bicyclists, cars, and trucks caused the processor to gobble up a grand total of 63 milliwatts."

Neural Net algorithms that would run very well on that kind of processor are usually very ideal for things like pattern recognition. They're often very bad at other things; so I wouldn't jump to too many conclusions.
 

T9D

Diamond Member
Dec 1, 2001
5,320
6
0
Sounds perfect for future smartphones. Watch Dogs, anyone?

"To give some sense of its actual performance, the researchers got TrueNorth to do object recognition on a 30 frames-per-second, 400×240 video feed in real time. Picking out pedestrians, bicyclists, cars, and trucks caused the processor to gobble up a grand total of 63 milliwatts."

Wow using that little of energy it's like a persons own body could be the battery.....oh....wait
 

crashtech

Lifer
Jan 4, 2013
10,521
2,111
146
Wouldn't it be weird if this turns out to be the next bug thing, and IBM ends up back on top again?
 

irishScott

Lifer
Oct 10, 2006
21,568
3
0
Why do we always assume that a self-aware artificial intelligence would want to kill all humans? Maybe we would develop a symbiotic relationship with them.

More likely they'll be incapable of harming us due to hard-wiring. We can already brainwash regular humans to do just about anything, let alone something we build. Wouldn't have to be anything as crude as a kill switch either, we'd just eliminate any ability to feel pain, suffering, betrayal, etc. We could make the system as predictable as we'd like.

I think the whole "AI kills humans" thing is just due to our lack of knowledge of AI. We project our own consciousness onto a machine and ask what it would do. A more realistic perspective is the machine gets some very carefully tailored consciousness that naturally constricts it's thoughts to it's given task. Imagine a lower level animal like a dog who has perfect obedience but also massive technical analytical abilities. It's nothing nature would produce, which makes it hard to process, but it's certainly possible.
 

irishScott

Lifer
Oct 10, 2006
21,568
3
0
Wouldn't it be weird if this turns out to be the next bug thing, and IBM ends up back on top again?

I'm sure practical implementation is still a ways off, but it's quite possible that it is. The one thing truly stonewalling Minority Report-style stuff is bad pattern recognition. An indefinite number of factors have to be taken into account, basically we're talking massive decision trees and massive amounts of hardware to feed them properly. Given that neural networks excel at irregular pattern recognition, they could very well be the answer. I've also read that quantum computers could potentially be utilized in a similar fashion.
 

Red Squirrel

No Lifer
May 24, 2003
67,194
12,025
126
www.anyf.ca
It would not surprise me if some of these things are 10x more advanced and in use by the government already but are top secret. The companies like IBM may be the ones working on it and they're allowed to release some info on the state the tech was in 10 years ago, but lot of it is hush hush between them and the government.

/tinfoil hat.

No seriously though, would not really surprise me if the NSA already has stuff like this that is fully functional and being used. They can't possibly be looking at all of our data manually, it's all AI doing it.

The Spynet NSA Neuron Network. Has a nice ring to it.
 

irishScott

Lifer
Oct 10, 2006
21,568
3
0
It would not surprise me if some of these things are 10x more advanced and in use by the government already but are top secret. The companies like IBM may be the ones working on it and they're allowed to release some info on the state the tech was in 10 years ago, but lot of it is hush hush between them and the government.

/tinfoil hat.

No seriously though, would not really surprise me if the NSA already has stuff like this that is fully functional and being used. They can't possibly be looking at all of our data manually, it's all AI doing it.

The Spynet NSA Neuron Network. Has a nice ring to it.

You'd be surprised how awesome our current analysis techniques are. I'm no data mining expert, but it's come a long way in the last few years. Just look at the facebook graph search, and how much information you can glean from that. Now imagine if it was better and privacy settings didn't matter. No AI required.