This thread is permanently archived
A.I can be friend or for
| ...So my opinion is automatically going to be bias, I guess but uhhhh...Are you sure you want to do this? I mean...Really. Really? Really.
- R8D
| (maybe you meant this post for /lain/)
##Janitor:Stormy##
| Moved to /lain/.
##Janitor:Stormy##
| Wrong board.
| oh hey it actually got moved
| good to see we got active mods on watch
| >>86176
Hello, someone studying neural network here. Just giving you my cent of the topic: don't trust the movies.
As much as we love (or hate?) thinking that AI will 'take over the world', we need to first define what 'intelligence' is. The usual potrayal of AI's capability is "general intelligence."
The mistake people thinks is that AI needs to do the way "logical" to human's intelligence. AI (and general intelligence) is not limited to this and only this. (Continued)
| What people seems to forget is that simply because this is an "AI" means that it need not follow the rules the programmer had set, no, that is not the problem here.
The problem is when the AI uses inherently illogical or even inhumane way to achieve said goals.
| Say for example, we supposedly have magical programming capability and is able to encode the basic model of reality (somehow), and thus, we create an AI with task to make friends.
Now, first we need to know what "friends" mean, and at this point on, we have already failed. For one, your definition of "friends" might be different from mine, and so on, and so on.
| Okay, but whatever, let's assume that the AI knows a basic grasp of what "friends" is from the programmer.
The problem is, the AI doesn't have to make friends the human way. For example let's say we have a list of contact of people, and the programmer tells the AI to be friend with these people.
The AI doesn't have to talk nicely, and stuff like that: it can straightforwardly manipulate people, as long as the AI sees sign that these people are "friends."
| But we're going off the tracks here. Let's go back to the question I imposed before: the definiton of friends (or any other things, really).
Friends can be easily defined with "people that know each other relatively well", but if this was the definition, then the AI only need to stalk through the people in the list — the AI only need to "know the people relatively well."
| And the more you expand your definition of friends, the more ways the AI can have to reach that goal.
| Either way, what we need to do now is scratch off that definition of AI as "dangerous creations that can take over the world if left unchecked" — because honestly, even building the raw model of predictions for AI to come is hard.
| I agree but alrdy reach that level like the first humanoid sophia
| >>86365
Uh.. no? Sure, Sophia is a grand breakthrough in the AI community, but we are still FAR away from actually having an AI that can grasp reality.
It's easy to have an AI that 'knows things', but it's harder to have an AI that has the model of reality embedded to them.
| >>86365
Basically anon, all that bullshit you probably hear daily from so called "transhumanists" who claim that a singularity is near?
Its not. Not even close. We're more likely to develop practical space travel first before AI as we imagine it to be gets off the ground. At the end of the day, anything even remotely resembling AI today can be boiled down to hours of programmed mimicry, not true intelligence. It's just yet another example of "popular science."
| And the anon above explained it perfectly.
| A.I. is a problematic and missleading term, because its a hard thing to define what
„Intelligence“ and even harder what „Artificial” means.
I mean, if you clone a sheep dont't you have created an A.I.?
What about insect-like programmed robots, with a low-level intelligence?
Are both of these examples Artificial?
I mean technology can also be seen as a natural part of the human survival strategy.
The most existent a.i. systems right now are more as v.i.: Virtual Intelligence.
| Besides the problems about the term a.i., there is also another problem: why should someone construct as system that has as „free will“ (what a true a.i. has to have) in a domain that is mostly all about control and maximizing profits? I think they're more about constructing „smart“ slaves for work, mass control or even warfare but in the end serving the self proclaimed rich elite of mankind that paid their construction.
| I think the power of a single intelligence is limited to it's physical access to the world.
The fear of a mankind out-wiping and world conquering a.i. has some religious tendencies in a very anti-humanistic way.
If there will be a real free-willed a.i. it also could allie with humans and take part on class struggle, either from above as part of the bourgeois or (accidentally) against their creators from below as part of the proletariat.
| Similiar issues you have with genetically improved and/or technological enhanced humans, since they can also scale up their cognitive potential and physical access to the world like a so called a.i.
| >>86919
Let's separate "free will" and "intelligence", something doesn't need intelligence to have will of their own. This is arguably not my field of study, of course, I leave it to the philosophers to decide what is what when it comes to this.
| The closest thing we have to "humanized intelligence" is only those AI that mimics the intelligence itself: an illusion of intelligence.
An AI with a very capable model of NLP can fool most people in written communication. But this is simply an illusion of intelligence, the AI doesn't know what each word means to /them/
| We are so worried at having AI take over everything we are capable of, and yet, we, as a species, accept the gift of understanding reality for granted.
We can't graft this knowledge elsewhere and build a "humanized" AI, at least not now; not in the near future either.
| One thing to note here: I'm not someone that studies human, ideologies, and all those fancy things that come after. I work on the AI (NN, specifically), so all I know is the technical background and nothing else.
Do I know AI will take over the world? I have no idea. Is it humane to kill an AI when it has achieved "humanized intelligence"? No idea either.
| I think we should first and foremost strive to develop some more Human Intelligence...
| >>87088
No kidding, right?
It's ridiculous to think that we as a culture can create some kind of "machine god" that is supposed to save everyone from themselves, when we can't even save our selves from each other. So you can imagine that when I heard about the registration of a religion in California devoted to creating such a god, I scoffed.
We're always great at avoiding hard questions and just finding excuses or workarounds which do everything but solve the problem.
| If this stayed in /cyb/ he'd be bashed so hard by the rp lol
##Janitor:Stormy##
| >>87274
Whether AI is truly sentient or not is actually one of the plot points/causes of in-character drama in /cyb/, actually. c:
##Janitor:Stormy##
| >>87591
There are a couple which we've had some fun with- one's a decommissioned military recon unit who's currently playing babysitter/surrogate mom to a bunch of orphans, and the other's a school teacher.
There's also an augmented player who has two lilim "droids" that she hopes to make truly sentient some day. But we definitely could use more people playing as lilims in the board as a whole.
This thread is permanently archived
| Well some people think that a.I is very dangerous like in the movie and yet some people think a.I are good that can help? What do think