Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[Reelya]

Yeah, but my point is that they're not similar or alike, which is why I said they're not analogous. That was the entire point. You just bringing up the definition of an analogy doesn't change the meaning of what I said in the first place.

A number which is supposed to represent some complex quality is not analogous to the actual quality. My original point that you took exception to.

e.g. you then said "but they are analogous because an analogy has been made". But this is ... just complete rubbish. Things that aren't alike don't magically become alike just because someone says so.

[strainer]

It looks like you use analogous as a fancy word for 'similar', but it has a particular meaning. Ironically here its plain meaning eclipsed your assertion that: "this things numeric analog is not similar to the thing". Similar? - depends on what you mean, but analogous? - Analogs are certainly analogous - unless they are broken.

[xordae]

No AI we've created thus far exhibits any kind of life or sentience or self-awareness. We're lightyears away from a true '13th Floor' or 'World on a Wire' style simulation. The AI in Dwarf Fortress for example has no ability to self-improve, mutate or otherwise do something unexpected. It's just a beautiful (and very simple, when you get down to it) numbers game that you can watch unfold, manipulate and poke.

I think we as a people have a pretty big problem of misattributing intelligence and awareness. We're frequently painting this horror picture of AI and evolving technology around the IoT as a problem, while thinking of ourselves as biological machines instead of our awareness being a spiritual quality. We're also stubbornly refusing to award sentience to the animal kingdom beyond maybe companion animals vs. the uncountable billions that we've delegated to a short life of suffering.

I didn't wanna put such a predictably lame spin on it, but this is an ethics question. Basically: There are great ethical problems to be overcome in our time. AI isn't one of them (yet).

[GoblinCookie]

No AI we've created thus far exhibits any kind of life or sentience or self-awareness. We're lightyears away from a true '13th Floor' or 'World on a Wire' style simulation. The AI in Dwarf Fortress for example has no ability to self-improve, mutate or otherwise do something unexpected. It's just a beautiful (and very simple, when you get down to it) numbers game that you can watch unfold, manipulate and poke.

I think we as a people have a pretty big problem of misattributing intelligence and awareness. We're frequently painting this horror picture of AI and evolving technology around the IoT as a problem, while thinking of ourselves as biological machines instead of our awareness being a spiritual quality. We're also stubbornly refusing to award sentience to the animal kingdom beyond maybe companion animals vs. the uncountable billions that we've delegated to a short life of suffering.

I didn't wanna put such a predictably lame spin on it, but this is an ethics question. Basically: There are great ethical problems to be overcome in our time. AI isn't one of them (yet).

The problem here is that how would we ever know if we did?  How do we know if the thing we are dealing with is actually an intelligent being or simply a sufficiently advanced impression of one?

[KittyTac]

No AI we've created thus far exhibits any kind of life or sentience or self-awareness. We're lightyears away from a true '13th Floor' or 'World on a Wire' style simulation. The AI in Dwarf Fortress for example has no ability to self-improve, mutate or otherwise do something unexpected. It's just a beautiful (and very simple, when you get down to it) numbers game that you can watch unfold, manipulate and poke.

I think we as a people have a pretty big problem of misattributing intelligence and awareness. We're frequently painting this horror picture of AI and evolving technology around the IoT as a problem, while thinking of ourselves as biological machines instead of our awareness being a spiritual quality. We're also stubbornly refusing to award sentience to the animal kingdom beyond maybe companion animals vs. the uncountable billions that we've delegated to a short life of suffering.

I didn't wanna put such a predictably lame spin on it, but this is an ethics question. Basically: There are great ethical problems to be overcome in our time. AI isn't one of them (yet).

The problem here is that how would we ever know if we did?  How do we know if the thing we are dealing with is actually an intelligent being or simply a sufficiently advanced impression of one?

Analysing the source code. They are explicitly governed by simple processes.

[GoblinCookie]

Analysing the source code. They are explicitly governed by simple processes.

So are brains.

[KittyTac]

Analysing the source code. They are explicitly governed by simple processes.

So are brains.

Dwarves are an order of magnitude or two less complex than humans.

[GoblinCookie]

Dwarves are an order of magnitude or two less complex than humans.

You proposed analyzing the source code, which amounts to reducing the thing down to a simpler level.  The sum of the scripts makes the program, to break the program down you will end up with simpler and simpler scripts.  Brains are also like, the neuron alone is simply a cog in the machine but the sum of all the neurons is a person. 

[xordae]

Brains are also like, the neuron alone is simply a cog in the machine but the sum of all the neurons is a person.

Let's not forget that a theory is not a fact. And that even scientists can't agree on what makes a person. They may have figured out how memories are stored, how nerves grow and how injury to the brain results in personality changes, but when you get down to it - we know very little about the brain still. And whether or not it's the seat of consciousness.. about that, we know exactly zero.

[Reelya]

And that even scientists can't agree on what makes a person

Just not being able to agree on the exact definition of something doesn't mean that things aren't generally classifiable.

e.g. say nobody exactly agrees on what makes a planet. Does that mean that we might as well call your underpants a planet, as much as Venus? No, it does not. Even with a "fuzzy" definition that we can't quite explain in words, we clearly know when something isn't the thing in question. You don't have to know anything about how planets work to tell that something is (or could be) a planet, or just isn't.

Similarly, while the actual definition of what a person actual is might not be a solved issue, that doesn't mean we can't classify almost everything in the world into people/not-people categories pretty easily. And a couple of bytes in a computer memory isn't a person, even if you stick a label in the program that says "this is a person, honestly". The label is a bald-faced lie, honestly.

e.g. you can put a letter "T" in your game and tell people it's a tree. Is it an actual tree? No, it is not, and not having an in-depth knowledge of detailed tree-anatomy doesn't change that. Similarly putting a bit of arbitrary data in a game and calling it a person doesn't make it a real person, or anything remotely approaching a real person. And not knowing exactly how real people work doesn't make it any less certain that a scant few bytes of data in a computer's memory is not a real person.

[GoblinCookie]

Let's not forget that a theory is not a fact. And that even scientists can't agree on what makes a person. They may have figured out how memories are stored, how nerves grow and how injury to the brain results in personality changes, but when you get down to it - we know very little about the brain still. And whether or not it's the seat of consciousness.. about that, we know exactly zero.

Indeed, but it is fairly solid to say that consciousness has something to do with the brain, even though the mind and the brain are not the same thing in my opinion.  However if we are not talking about the consciousness side of the thing but essentially the mindless reflex side of the thing we avoid that philosophical question entirely.  The brain is something that does something, but if you tear the brain apart you do not arrive at anything that specifically does something, but instead we end up with lots of little things doing lots of little things the sum total of which adds up to doing something. 

A computer program works similarly, if you break it down you get a lot of little scripts.  There is no "this is a person" script for us to identify in order to tell if our prospective AI program is actually a real entity or just a regular machine.  This supports the general argument that it is impossible to know if you have genuinely created an AI which in turn supports my argument that the appearance of wrongdoing is what matters not whether you are actually doing any wrong objectively. 

[strainer]

There is no "this is a person" script for us to identify in order to tell

Here is an idea of a function which would seem to disable a program being truely sentient,  its a very common function that can be either of two distinctly different variants. The two are functionally interchangeable - swapping them should have no empiricallly discernible effects on the overall AIs character or performance, yet a metaphysical difference enables one kind to rule out the presence of 'true' sentience.

The function Im thinking about is random(). If random is not used, or is the seeded psuedorandom kind, then now matter how impressive the AIs output is entirely determined by its input. Include a hardware source of entropy for random(), and the AI can become unpredictable. The two versions shouldn't appear any different, just as we couldn't tell if DF worldgen is seeded by one method or another, yet one AI has a fixed response to its input history, the other AI has degrees of freedom which might approximate free will.   

[KittyTac]

The problem: it is obvious that dwarves are nowhere near as complex as humans. They do not actually "learn" anything, they do not actually "feel" anything, and they do not actually "think" about anything.

[Reelya]

Determinism and sentience aren't actually incompatible. e.g. it's perfectly possible to have the sense of free-will even in a deterministic universe.

e.g. if "you" make a decision, that decision is determined by your previous state. But you are the state. So there's no "external" force "making" you do what you didn't want to do. The confusion comes from the idea that something "external" forced you to act as you did. But it didn't, because you are the deterministic system. Determinism is in fact internal decision-making because the self and the system are not a duality, they're the same thing. There's also nothing special about humans in this. We're just biomachines who have feedback/sentience. There's no need to start creating new pseudo-science physics because we're uncomfortable with the idea that the existing laws of physics might pre-determine what we do. We're just not special enough in the universe to warrant that.

The idea that hooking up a "random" input source means free will is wrong. That's not freedom, that's being buffeted uncontrollably by whatever random fluctuations happen to occur. In a sense, that could even be said to be less free than just being a deterministic being who uses it's own state to decide how to act next.

[Egan_BW]

Without determinism, free will cannot exist~