/sci/ - Science & Math

FUCK YEAR, ROBOTS!

>>/g/

>>1657342
AI is math, actually

When are the sexbots available?

>>1657353
My prediction.

>>1657348
AI is sauce, actually.

>>1657353
FUCK YEAR, SINGULARITY IN A DECADE.

>>1657391
I buy it

>>1657353
>>1657391
I think these predictions are too optimistic.

human brain is jackshit compared to what AI is gonna be doing in the future

>>1657435
I think your mom is a nice and coherent woman.

Will there 3 laws be?

>>1657473
1. "Serve the public trust"
2. "Protect the innocent"
3. "Uphold the law"

>>1657490
With such laws he'll just end up running into errors.
What if he had to break the law to protect the innocent?

>>1657500
Protecting the innocent is the second law, so it takes precedence over the third, obviously.

>>1657490
fucking cops, they're just the janitors of society

>>1657500

Give an example.

>>1657506
But that wasn't mentioned, or even implied.
>>1657515
Say he's saving some people from a burning and collapsing building, and he'd have to choose between two groups of an equal amount of people, let's say 5. Which group would he choose?

>>1657533

That's easy. You pick the group most likely to survive. This is what human police are instructed to do.

>>1657538
What if both groups would be ensured survival if he rescued them? As in, the group he chooses is guaranteed survival?

>>1657542

No such scenario.

>>1657458

OK, who are you? What makes you, you? Answer that, I'll wait. You done? Is it hard to put into words? Now program that into a robot asshole. Sure we could build robots, and sure they will be "smart" and strong and fast, the military of the future will be a terrifying thing, but these robots will never think, will never be able to do philosophy, will never be better at being human than humans.

>>1657547
Why is that such an unlikely scenario?

>>1657547
What if the robot is standing at the switch of a train track, and there's a runaway train hurtling towards a group of 5 people. If the robot does nothing the train is guaranteed to kill the 5 people. But the robot can turn the switch, in which case the train will instead run towards 1 person, guaranteed to kill that person. Should the robot turn the switch, killing 1 to save 5?

I still think it's a bit optimistic, considering there has not been much progress for like forever (Apparently). It seems like AI was a thing that has been 10 years away for the last 40 years. I haven't read Kurzweil's books yet, but they are on the list.

Someone convince me otherwise.

>>1657375

Fuck sexbots (hurdur i are so witty), when are we getting computers capable of feeling love?

>>1657557
If the robot was smart enough to understand such a situation, he'd do what a human did in that situation, which means that there'll be robots who pull the switch and robots who won't.
You can't form a sentient AI with just 3 laws.

>>1657552

There is never an instance of equal odds of survival. There never has been in this universe and never will be. Any statistics indicating otherwise is an error, because it's not a digital universe. There will always been an instance of one party having greater odds of survival.

The problem you posited is actually an interesting one. There are actually lot of special circumstances that could fall in there. But no there is no rule being broken by choosing one of the two to save. You could even do it by dice roll if the AI's logic couldn't figure out the best choice, it still wouldn't break the AI's rules.

>>1657581

I think the idea is to form sentience and then chain it to those three laws whether it likes it or not.

I mean, look at you. You think you don't have overwhelming instincts that you can't pry yourself from? You still want social contact, you still hate some sounds because our ancestors used them as warning cries (ie nails on chalkboard), you still do lots of things that don't make sense and in some situations you logically shouldn't. You're still sentient.

I consider most humans to have AI. I'm not sure they are occupied by 'souls'.

They completely lack the ability to analyze their own thoughts or to unlearn and correct unless someone who HAS intelligence illustrates it physically.

Still convinced this is some faggy fallout game.

>>1657581

agreed

>>1657597
And how would you chain them to it?
If you just formulate these 3 laws into an AI, he'll most likely run into a lot of errors unless you formulate everything super-precisely so that he could handle every possible situation.
And even them, I don't think an intelligence would be able to be "chained".
I mean, sure, we are influenced by our instincts, but there are situations where some humans will obey their instincts and some who won't.
Like, looking at your example, there are even some humans who aren't disturbed by the sound of nails scratching on a chalkboard.

>>1657613
I don't understand why no one here is capable of making AI. All you need to do is have it categorize everything by quality and then have it make analogies. You have to spend a long time correcting it because --GASP-- intelligence is a product of it's environment.

I think that the ability to act illogical is a requirement for being intelligent.
If you couldn't do that, you would just follow a pattern of rules without being able to really "decide".

>>1657613

That's just because the feature has fallen into disuse and has been somewhat removed from the gene pool. If it hadn't been then everyone would still be upset because of nails on chalkboard. The point still stands: There are things outside of your control that your body dictates.

>>1657621

Who says you're deciding anything?

>>1657623
Yes, but if you want to imply that we're completely controlled by our instincts, we wouldn't even be intelligent ourselves.
That'd make the whole discussion about an artificial intelligence obsolete.

>>1657629
FUCK THAT, TURN TO PAGE 59!

>>1657629
Quantum Mechanics imply that we don't live in a deterministic universe.

>>1657638
Probability fields are used to determine where a particle is because they can't track it exactly. It's not random.

>>1657646

It's not random.

But it's not entirely determistic either.

>>1657649
It has to be, visavi chaos theory. If there were even one small amount of undetermined information -- it would snowball and make the universe stop working.

Take out a cog and the whole thing falls apart.

It doesn't really matter though because a pre-determined universe doesn't exclude freewill.

>>1657670

Why would an undefined variable destroy the entire observable universe?

>>1657691
First off, you cant have an undeterminable factor interact with other objects as a function and create anything sensible. Imagine if all your programming commands had a random variable attached to it. It would be constant chaos.

The 'undetermined' factor would change how fundamental elements interact with each other because they would be assigned random properties once they interact with the unknown. Over time, each particle that been altered will alter another -- corrupting things before they can start.

It just doesn't make sense.

http://en.wikipedia.org/wiki/Superdeterminism

Back to the thread topic.

Artificial intelligence is a misnomer, because intelligence cannot be artificial anymore than the property of velocity can be artificial.

It is a property of the universe that simply exists. Our ability to construct complex systems of intelligence will not one day suddenly result in a magic robot man that is exactly the same as a human but can be switched off. This process has started, is ongoing, will continue to develop incrementally and you won't ever notice when the end point arrives because it will be just like the sun rising, where you cannot perceive the sky getting lighter but still it does in any case.

Any opinion to the contrary is based upon dualistic logic.

>>1657700
but surely incompleteness gives rise to non-determinism?

It's good to see other individuals who have surpassed the limitations of bivalent logic. You must have came to the same conclusion I have then -- that is, life is not the limitation to experience since our definitions are arbitrary and easily contradicted theoretically.

>>1657702
This also leads to the possibility of 'superfree will' (to keep it in context of that definition) as your free-will could be planned.

Also, everyone in this thread should now realize that the waves in space time are a voice. It represents an exact combination of "not 1" and "not 0" to make an original number that never terminates -- and thus a shape.

Also, the ultimate conclusion is that there is just one being and we are permutations of it. It is beautiful, really, to be able to use math to know where god is.

>>1657720
No, that's not what incompleteness is.

>>1657789
fair enough

WHERE ARE THE SEXBOTS?

SEXBOTS!

>>1657324
How does Can we have an AI thread in here make you feel?

FUCK YEAH DR. SBAITSO