Counting to 1,000,000

[Brisky]

171,512

Except we can do stuff.

That's something a computer will never be capable off. It'll only be capable of doing as it's been programmed to do.

[HellEye]

171,513

Well
Until we get to AI that will take over the world.

Though we're basically already there with AIs telling you what to watch on YouTube and what to see on Google

[Cheza]

171,514


Easily thwarted!

Just ask an AI what a woman wants to eat!

[HellEye]

171,515

I mean it won't get it right I'm sure but it can give you a good average.

[Brisky]

171,516

Yes, perhaps, but we're still only at the point where the AI suggests youtube videos because it has been set up to do so.
It doesn't really make decisions, it takes inputs and produces a results based on calculations or comparison.

It's getting more interesting nowadays with AI learning, and improving itself overtime for a certain task, yet still it is doing this towards the task it has been set up to criticize itself towards.

[HellEye]

171,517

I mean
I'm pretty sure if you had enough computing power and was like "AI, be a human" and then got millions of scenarios and millions of answers to each one, you could model an average human decision making.

[Brisky]

171,518

But can you make "be a human" a criteria that a computer can understand so it can work towards this?

And even if, it's still but a simulation of human decision taking. It's "making decisions" because that's what it has been set up to do. And if you understand how it has been programmed to do this, you can likely predict what decisions it will make.

[HellEye]

171,519

Well you can't make "do well in Mario cart" a criteria either.
It can only be "try to minimise error", that's how they all work.
And in some sense that's how humans work as well.
We do what we think is best.

That's why I said it would have to be massive. And you'd need many different scenarios, so it wouldn't just be like "run through a red light", it would be conversations and eating habits and all that sort of stuff that humans do.

And yes it would probably come out to be literally your average Joe living in the suburban house with his wife two kids and a dog, but I'm sure with enough tech you could twist it around.


Now I didn't say it's possible now.
I'm just saying it is possible.
Neural networks are how actual brains work as well, you just need a thousand interconnected massive networks and a lot of learning data. Or generic learning with a ton of random junk that just tries to match the given answers somewhat closely.
So it's not possible now but could be in the future.

[Brisky]

171,520

But the thing I'm curious about is: It'll undoubtedly be possible to simulate the behavior of life using AI, but can it become life?

Yes, we essentially function the same as a computer but we (as difficult as it is to philosophically understand what that term refers to) eventually do things because we enjoy them, or react to not enjoying things.
That's where I'm trying to go. We make decisions because we want to, more or less.
And can AI reach this? Or will it, however advanced it may one day be, remain stuck in a position of doing things because it has been programmed to do so?

[HellEye]

171,521

I think with enough data and weird behaviours being taught it could have random "mood swings" or something.
But no, I doubt it'll ever be "conscious", even if it would say it is.

That being said, it could still have a random destructive "mood swing" and nuke everything.
Or commit suicide

[Brisky]

171,522

Yeah, probably...

And there has actually been this one Japanese chatbot that got suicidal after having been "red-pilled".

But it is as you say, it probably never felt suicidal or unhappy, but it was just a thing it's programming picked up to say during the chat sessions.

[HellEye]

171,523

Yeah.

I mean, maybe if we really go deep on AI research we will be able to create something that's really conscious.
If our brain is conscious there's really no reason some wires couldn't if wired properly.

But we're ways off that. In the next century best we'll get is closely simulating human behaviour without actually making it "alive".
At least that's my thought.

[Brisky]

171,524

Yeah, but we'll probably learn a lot in that process!

[HellEye]

171,525

Definitely.

So I randomly got some hummus the other day
And my god it's good

[Brisky]

171,526

I think I tried humus once and hated it, but glad you don't I guess!