don't be pedantic, you obviously know what i mean

It's a point worth making, though.

You're implying we have control over whether or not a certain intelligence--regardless of neural/computational complexity--develops sentience.

Quote from: Pendulate on June 21, 2015, 03:18:38 PM

If sentience in AI were to happen, it would almost certainly be by accident.

we can't give things by accident?

come on

don't be pedantic, you obviously know what i mean

No, I took you to mean that we'd intentionally endow a machine with sentience.

You know, I think the closest you can come to is an intelligence that's allowed to modify itself. Which, we already build rudimentry versions of. Little processors that can change their own coding to better suit certain aspects of things that they know.

If you create a particularily powerful intelligence that has the capability to alter it's own programming, then I don't think it would be too hard to eventually, essentially, stumble into the realm of being considered a sentient entity.

i mean, i would just call that agency, not sentience

for me, the defining factors of sentience involve a lot more than just agency

can it think? can it feel?
can it communicate its ideas?
can it get offended? can it get hurt?
can it suffer?

does it have interests/disinterests?
does it have a personality?

that's what makes sentience to me, and if you disagree with that, then i'm not talking about the same thing as you

Perhaps although it depends on the situation.

i mean, i would just call that agency, not sentience

You don't think agency is a vital component of sentience?

If you think an AI needs to think and feel in the same ways as we do in order to be deemed sentient, then no, we probably won't ever have sentient AI.

If it just needs to be able to think for itself, and move beyond its man-made programming, then it's well within the realm of possibility.

Quote from: Sandtrap on June 21, 2015, 03:38:39 PM

You know, I think the closest you can come to is an intelligence that's allowed to modify itself. Which, we already build rudimentry versions of. Little processors that can change their own coding to better suit certain aspects of things that they know.

If you create a particularily powerful intelligence that has the capability to alter it's own programming, then I don't think it would be too hard to eventually, essentially, stumble into the realm of being considered a sentient entity.

i mean, i would just call that agency, not sentience

for me, the defining factors of sentience involve a lot more than just agency

can it think? can it feel?
can it communicate its ideas?
can it get offended? can it get hurt?
can it suffer?

does it have interests/disinterests?
does it have a personality?

that's what makes sentience to me, and if you disagree with that, then i'm not talking about the same thing as you

I certainly don't disagree with that.

And, there's no real feasible reason why an intelligence that was able to self modify itself to adapt to new encounters couldn't reach that. Let's say you build an intelligence whose core revolves around public interactions with people.

That's a lot of raw data and a lot of conflicting encounters because people and their personalities are all varied. Eventually, the intelligence could learn how to filter through all of that data and a distinct, unique personality may emerge.

Now, that's not going so far as saying it would be sentient. But, then again, who is to say that with exposure to people and their emotions, it couldn't adopt the usage or simulation of them to better interact with people?

Knowing that a smile isn't hostile, and thus choosing to smile at a certain time for the benefit of the person it is interacting with?

You might say that would be parroting.

But, kids parrot their parents as well. And some people, do the things that they do, without really knowing why. Without questioning why they do it when they inherited their attitude and actions from exposure to their parents.

I certainly think that an intelligence that was allowed to modify itself could easily fall into the realm of sentience.

That's a poor definition of sentience, though.

it's actually the best definition ever

present me with a better one

I never said it wasn't a "vital component"--I'm saying that all it is is a component. It's not the only thing that defines sentience.

Well, I'd say it's enough of a factor to define it for the purpose of this discussion.

Unless you want to coin a new term for computer sentience (intelience? Swidt) then I don't think this is a misuse.

I'd also take this to mean that you'd support sentient AI, in this case? Assuming they can't suffer in ways that we can, and they can perform complex tasks in short amounts of time?

present me with a better one

Well we could just use the actual definition, which is the ability to experience subjectively.

Which doesn't at all require any similarity to human experience besides some aspect of subjectivity and, presumably, self-awareness.

I would say developing a true artificial intelligence would be the single most important discovery in human history; all others would be meaningless by comparison except in their contribution to AI. It would mean we understand how our minds work and what actually leads to sapience. Beyond a mere function of processing we'd be able to make the existential into the scientific. Questions of existence and origin would leave the realm of pseudoscience and become experimental.

Artificial intelligence isn't sought by making the most efficient computer, it's by unraveling the mystery of consciousness and reason. A true AI would mean we've finally prioritized knowledge over utility, and that would be a turning point of our species.

Thank you.

Improving the species is and should, always be the goal.

because that would contradict you, and help my definition instead

No, it wouldn't.

no one's strictly talking about human experience

That seemed to be what Pendulate was implying when you used his consideration to demonstrate your point.

it's my exact definition of sentience, though

And you're yet to demonstrate why artificial intelligences would be excluded from the ability to perceive, feel or be self-aware.

i'm of course referring to all sentient life

See above.

We have very little evidence as to how other animals feel and percieve things

I'm not entirely sure this is correct.

Quote from: Meta Cognition on June 21, 2015, 04:28:31 PM

And you're yet to demonstrate why artificial intelligences would be excluded from the ability to perceive, feel or be self-aware.

it's not my job to

that which can be presented w/o evidence can be dismissed without evidence

the onus is on the person who's making the positive assertion--and i'm making no positive assertions
i know you don't agree with that logic, because it's inconvenient, but

Except you are.

You've quite clearly claimed that sentience is not the result of complex neural/computational networks, which is currently our best hypothesis for the existence of sentience/conciousness.

You've made a positive claim which draws the distinct conclusion that machines are exempt from such sentience, and I'm waiting for you to justify it.