Will we ever have true AI?

will we ever have true AI?
all so-called artificial intelligence to date is just VI (virtual intelligence) just things that seem to be thinking or alive. it's just tricks and illusions and lies.
when will computers actually become sentient?

Other urls found in this thread:

nature.com/news/fragment-of-rat-brain-simulated-in-supercomputer-1.18536
twitter.com/AnonBabble

>things that seem to be thinking or alive
>implying this doesn't apply to humans

this is Sup Forums, not /x/

no

Exactly, you think we are imbued with some mystic power?

i shit posted over 20 pounds of shit on Sup Forums today.

We'll have AI as soon as we need it for better sexbots.

If you could put on a device that measured all of your brains chemical reactions, it would be possible to tell exactly anything you were to do, before you did it.
So, how are we any less deterministic than current ai? We are given inputs and produce an environment changing output.

I think humanity will end before we achieve such technology

We'll get closer as humanity gains a larger number of Lisp users.

nature.com/news/fragment-of-rat-brain-simulated-in-supercomputer-1.18536
Do you think that rat existed with the same conscious level that any normal best would possess, be it solely for a split second?
>

Give me a definition of "true AI" and i'll make your ai for you OP.

Or at least describe what an executable file called "True AI.exe" would do once you open it.

AI is a moving target so no.

the pinnacle of AI as most would see it: Data from Star Trek TNG

just assume it has a humanoid form.
it should be able to understand and converse with humans via verbal, written, body language

it should be able to empathize
it would need to be able to learn new things via any communications and environmental factors

it should have it's own personality and therefore it's own interests, likes dislikes opinions, etc...

it should be able to become friends with people
it should be able to fall in love, even hate, hold a job, have hobbies

it would be able to distinguish individual persons...
also see: Bicentennial Man
the list goes on, but limited space here

So basically you want a humanoid robot that can do everything a human does?

AI is not about robotics, its about software. Try again user.

And don't say "hurr i want a smart chat bot".

Of course we do. It's called human consciousness. Human consciousness lets us speak and comprehend words and emotions.

consciousness is a term people use when they can't figure out where their thought originate from

look dumb dumb, i said the pinnacle would be embodied in a humanoid form. software is the key, duh! - but yes, the hardware is what makes the software work and is the way the software would interact with the world.

build it in the fuck cloud for all i care. put it on a microSD and we can futz around with it on RaspPi units with arduinos and shit.

What does it do though as software? What exactly is it?

Focus on the software.

I have a more interesting question for you, Sup Forums.

Everyone not retarded will admit that a simulation of a human brain could "genuinely" think and feel, if hooked up appropriately. Or at least, no test could tell the difference, so it is the same as far as science is concerned.

Going further, you could have thinking beings in a simulated universe. But what if, instead of stimulating a universe, you merely played back each the recorded states of a previously simulated universe? In a deterministic simulation, an observer outside the simulation can't tell the difference, if all they get to do is look at the states of the simulated universe over time. Of course a "real" simulation would respond to changes if you altered something in the simulated universe, while the recording could not. But if you chose not to alter anything, how is playing back a recording different from re-running a deterministic simulation from the same seed?

If you argue that it wouldn't matter, there's perception of consciousness in both the re-run and the re-play, then what if you printed out all the recorded states into flash cards, and held them up one by one? What if you just left then in a stack, or on disk?

From where does the perception arise?

see this:
and ignore the form parts.

Is 1+1 = 2 true even if nobody is calculating it?

The answer is yes. Same goes for anything more complex.

There need not be an observer for a universe to exist.

>it should be able to understand and converse with humans via verbal, written, body language
Okay so you're already feeding it data here. What should the data be in the form of? For text, Ascii? UTF-8?
For audio just a digital audio stream? For body language a picture? Can there be multiple sources?
What exactly do you mean by "understand"?

>it should be able to empathize
How do you determine if an ai is empathizing? What does it actually DO?

>it would need to be able to learn new things via any communications and environmental factors
Again, you're feeding it data. I want you to define clearly what kinds of data it takes and what kinds of data it outputs.

>it should have it's own personality and therefore it's own interests, likes dislikes opinions, etc...
>it should be able to become friends with people
>it should be able to fall in love, even hate, hold a job, have hobbies
See above. How do you determine personality? What does it DO?

>it would be able to distinguish individual persons...
Already possible with current ai

Also there's two questions to consider here. Determinism and time discreetness.

You're saying there's discrete states that follow one another and you can "freeze" at one point?
But is there really? How much time passes between two states? Any finite number you come up with, i can come up with a smaller one.
It only works if the universe is completely discrete in space and time.

As for universes that are not deterministic, you can think of them as multiple possible universes where some just happen to have the same beginning and are split at a random event.
A non-deterministic universe and a stream of random data equals a deterministic universe. Every time a random event needs to be resolved, you pull data from the random stream.
So basically you can think of a stream of random data to be a part of a specific deterministic universe.
And from any deterministic universe doesn't need an observer to exist in itself.

basically build a human mind

data in: what it sees and hears, touches, smells, tastes
then it would need to prioritize/categorize that information
ex: my field of view has maybe 100 items, and most have some sub-items, i don't have any need to focus on my camera/tripod to my left. i don't give second thought to my mousepad. my keyboard has 100+ keys and all but one at a time are irrelevant. but at some level i am aware of all these things when all i am focusing on is my monitor and what i want to say to convey what true AI should be.

empathy: any non autists should be able to console someone grieving, add a pun with folks are joking, not bother someone who is busy with something important.

I'm current working on it. The problem is that framework philosophy isn't being applied to self improving code.

There's obviously something different between humans and animals. That thing is consciousness and the ability to make long term decisions.

>what it sees and hears, touches, smells, tastes
it being a humanoid robot?

You see the problem user? You want a humanoid robot that acts like a human.
I want you to focus purely on software. The AI. General intelligence.
Or are you saying an AI needs a body or its not an AI at all?

>ex: my field of view has maybe 100 items, and most have some sub-items, i don't have any need to focus on my camera/tripod to my left. i don't give second thought to my mousepad. my keyboard has 100+ keys and all but one at a time are irrelevant. but at some level i am aware of all these things when all i am focusing on is my monitor and what i want to say to convey what true AI should be.
So basically image labeling. Already sort of exists.

>empathy: any non autists should be able to console someone grieving, add a pun with folks are joking, not bother someone who is busy with something important.
How? With text? Voice?

>long term decisions.
you mean like birds deciding to migrate for a whole season?

i'm telling you user, consciousness is to psychology's as aether is to physics

shit, if you're asking for all the pseudo code, i could just learn a language and write it myself.

all a computer does is process binary data.
the programmer needs to write the computer code that represents the real world.

if you wanna go super simple, pajeet, err i mean, code monkey, then try this on:

setup just one part of the whole. maybe facial recognition paired with a simple game.

the better one does at the game, the more points he earns.

AI assigns points earned to faces.
recognizes your face checks your score and sees you suck. so in response to you the AI acts more mean to you. and it loves me, answers my queries, offers helpful tips, would choose to defend me over you in a fight.

FFS forget the form. for all i care it could be a 3D rendering of a human in a modded minecraft or WoW world. - input is keyboard or voice recognition, out put is text or speakers.