The AI Revolution: Road to Superintelligence

[Molly]

Sitting outside of Uni, having a smoking break, I'm actually wondering what AI even means. At which point becomes something AI?
Commonly we use it even for game NPCs. Is that already AI?
How many decision trees do you need to qualify for AI?
Does it need self awareness?
What about consciousness?
How about the ability to learn?
Self upgrade?
Matrix anyone?

Hard to discuss something without a proper base line.

[Leshma]

It needs self awareness, can't be called AI without it. Which kinda means only true AI will be rogue AI. Something we probably won't ever allow, even if we come to that point where it could happen.

Gaming NPC behavior scripts aren't AI... closest to AI you can find in Black&White monsters (tiger, cow and monkey), because it is a work of actual AI researcher, where he gave his best.

Synergy is the future, not war for prevalence between humans and AI like in movies. That war would be lost at the very beginning. It is already happening, we are being upgraded on daily basis, our habits changed by computer technology. Many people already have artificial materials in their bodies, transitioning to fashion artificial parts won't be a big issue. Next will be augmentation of organs and other systems inside our bodies. Trans-humanism is the one of the few plausible scenarios that originate from sci-fi stories and the most likely one to happen imho.

[Xant]

Sitting outside of Uni, having a smoking break, I'm actually wondering what AI even means. At which point becomes something AI?
Commonly we use it even for game NPCs. Is that already AI?
How many decision trees do you need to qualify for AI?
Does it need self awareness?
What about consciousness?
How about the ability to learn?
Self upgrade?
Matrix anyone?

Hard to discuss something without a proper base line.

https://intelligence.org/2013/08/11/what-is-agi/

[Molly]

So there actually is no definition. Just several suggestion on how you could define one.
I imagined that it would be like that and I guess it's no surprise that it is such a cloudy thing.

In 1958 a suggestion was that a chess program could be already considered a AI and we're already pass that for nearly 2 decades.
The crux of many "new" research fields really. Everything is still in motion, dynamic and changing constantly. Sometimes due the complexity of the field, sometimes due to the stubbornness of the people in the field

[cmp]

Note the distinction between AI and AGI.

[Molly]

Random blog on the difference...

http://interintelligence.blogspot.de/2010/01/difference-between-ai-and-agi.html

Seemed a good enough quick explanation for me :x

[Kafein]

Haven't read any of the links but to me the definition of "AI" as employed by people in the field is pretty large. It's not even all that related to what the program really does but rather to the problem. When you create a program that plays a game you will always call it "AI" even if the strategy is simple. If you're building robots or software agents that behave independently of each other then it's "AI". When you have a classification problem like image recognition or spam filtering, you'll call it "AI". If you try to do market predictions or polynomial regression then it's not "AI". If it's some form of combinatorial problem but isn't playing a game then it's not "AI" either.

The nuances of vocabulary are really stupid and arbitrary. Ultimately though, there is no fundamental difference between AI software and any other information processing system. An IP stack is AI just as much as Google Car. You can make the case that one is deterministic and the other is not, but even that would be both wrong and an irrelevant detail.

[Dooz]

Obama issues executive order to create the world's first exaflop supercomputer, which can mimic the human brain

http://www.wired.com/2015/07/obama-supercomputing/

[Xant]

Amazing, considering the human brain is still a mystery to us.

Too bad that's not at all what the article says, retard.

[Gravoth_iii]

Obama issues executive order to create the world's first exaflop supercomputer, which can mimic the human brain

http://www.wired.com/2015/07/obama-supercomputing/

What? Did you actually read the article?

[Dooz]

copy pasted title from reddit before i did

[Kafein]

copy pasted title from reddit before i did

Wow, that's almost as much work ethics as an Ars Technica journalist.

[Xant]

Hasn't it become clear to you yet that Dooz understands literally nothing about the subject, he's just parroting and copypasting stuff he saw on some mainstream (AI!!! terminator picture!!!) site.

[Leshma]

That is probably supercomputer that will use newest nVidia Tesla cards based on Pascal architecture, connected with nVlink.

[Dooz]

Wow, that's almost as much work ethics as an Ars Technica journalist.

oops i forgot i'm a journalist getting paid for this and that i have any obligation to you people, sorry!