Would you date a machine?

[Asta Kask]

Is that an answer to or a context for your question?

"At a certain level of analysis"? It's context. At the level of individual elementary particles, there are no such things as pressure, temperature or an arrow of time. But as we get more and more particles these concepts become more and more useful. In the same way, an AI or a human could be deterministic machines at one level of analysis, but not at another. When we collect enough cells, we get complex patterns and phenomena such as free will become important. And I'm wondering if people are looking at AIs as deterministic machines and humans as free-willed subjects, and that is the reason for the stubborn refusal to date an AI.

But I also have a fever and one hell of a cold, so this may just be me rambling.

[enderlord99]

Humans transmit force. That's how we do just about everything.

That's true. I wouldn't date one.

[Frozen_Feet]

In what sense are humans not machines? At a certain level of analysis.

There's a colloquial definition of "machine" which excludes living biological creatures. Most people, when you say "machine", will think of a mechanical or electronical device made of dead parts. This is why I, for example, assumed the "machine" or AI I would be dating would be a computer program, rather than a biological computer or something akin to a Drakon from Geneforge series.

As a sidenote, the Geneforge series makes a pretty good study of the whole "is creation of AIs / intelligent machines ethical?" question. You just have to forget the colloquial definition of machine and accept that living biological things are, indeed, machines.

[TheThan]

That depends:
Sexy fembot? Sure (as long as it's not trying to kill everyone or is obsessed with me that is)
Ipad? Nope.

[noparlpf]

In what sense are humans not machines? At a certain level of analysis.

"A machine is a tool containing one or more parts that uses energy to perform an intended action. Machines are usually powered by mechanical, chemical, thermal, or electrical means."

I contain lots of parts. I use energy and perform actions. I'm powered by chemical means.

"A device that directs and controls energy, often in the form of movement or electricity, to produce a certain effect."

I direct and control chemical and electrochemical potential energy as well as kinetic energy.

"a piece of equipment with moving parts that does work when it is given power"
Full definition: "1e (1): an assemblage of parts that transmit forces, motion, and energy one to another in a predetermined manner"
"2a: a living organism or one of its functional systems"

Well, I have lots of moving parts at both the macro and micro levels. They all require power to function. I am also an assemblage of many, many small parts that convert electrochemical potential energy, chemical energy, and kinetic energy back and forth, giving off thermal energy as a byproduct. I am also a living organism by most definitions, which as I previously explained can also include many inorganic and artificial systems.

And just for fun:

Originally derived from PIE root *magh- "to be able, to have power" (cognates: Old Church Slavonic mogo "to be able," Old English mæg "I can;" see may (v.))

[Coidzor]

I am also a living organism by most definitions, which as I previously explained can also include many inorganic and artificial systems.

Which is still not going to get you anywhere when what you're up against is a rejection of expanding the definitions in that way.

[noparlpf]

Which is still not going to get you anywhere when what you're up against is a rejection of expanding the definitions in that way.

I'm not even expanding the definitions. I pulled out the textbook definition of "life" and explained how it applies equally well to organic and inorganic machines.

[Coidzor]

I'm not even expanding the definitions. I pulled out the textbook definition of "life" and explained how it applies equally well to organic and inorganic machines.

By arguing that we're machines in such a way in such a context as this in order to paint an aversion to romantic entanglements with AI as bigoted, you are in fact doing so.

Edit: And if that's not where you were going, then I am well and truly lost as to what you're really on about here.

[noparlpf]

I would argue that refusing to date a sapient AI on the principle that they are "not a person" or "only a machine" is, if not bigoted, at least uninformed. Personhood is not limited to humans (even on Earth); that's a very anthropocentric viewpoint. And humans are only machines. Refusal to date a sapient AI on the basis of lack of physical attraction towards their chassis or lack of interest in their personal qualities is, on the other hand, perfectly acceptable.

[Murska]

If you want to get hands-on experience on how this feels like, download DOSBox and play some old MS-DOS games on a modern computer.

I've done that. It's quite entertaining. Try the original X-COM (though you can get it working without DOSBox nowadays) or some of the old Sid Meier games. But I don't really think the comparison is that apt - it's not like me having or not having a modern computer changes the quality of the old games either way. The old games were just as bad or good back when they were made as they are now.

*snip*

Well, I would assume that trying to accurately simulate a human mind is not the way an actual AI would be built - it seems ineffective. Assume people understand intelligence enough to actually build an AGI in the hypothetical future that follows, that is created with an architecture that makes use of the properties of the hardware and is somewhat efficient, capable of self-improvement and starts at roughly human capacity.

Now, going back in the history of life, it pretty much started at some things on Earth, maybe in some pool somewhere, happening to begin self-replication. This had the result of increasing the relative proportion of these things in the area. And as these things replicated, sometimes errors happened. Most of these died, but some could still continue replicating. Those who were better at it soon became relatively more common - evolution began to happen. At some point much later inventions such as the DNA and sexual reproduction came along, optimizations found by evolution's slow and primitive optimization algorithm that searches the neighbourhood of the current position in design-space and latches on to more efficient designs that it finds, continuing the process.

Much, much later, there were various vastly more complex forms of life, all evolved by the same slow, inefficient, stupid algorithm. Humans came along. With enough intelligence, with language and culture to transmit knowledge, they began a new epoch in the history of life. The search algorithm for optimization changes. And, due to the way evolution has created humans, humans themselves do not optimize only for inclusive genetic fitness. Humans are capable of searching the design-space for better, more optimized designs by intuition and extrapolation, creating models of the world that let them entirely skip vast areas of possible designs that seem extremely unlikely to work and focus on those, much smaller regions where previous experience and knowledge of the world indicates you might find good solutions.

Now, if a human has an actual problem, there's a bunch of levels that do work towards solving it. Evolution designed brains. Brains invented knowledge on the abstractions surrounding the problem - how the physical laws of the world work. Then, using this knowledge that was invented by other humans long ago, another human who has learned of this knowledge through the transmission of ideas can apply it to create an object-level solution to the problem, like a new tool. This sort of development is slow, but it is vastly superior to the speed of evolution trying to design a tool to solve that same problem.

At times, human brains invented better meta-knowledge solutions. The best example would be the scientific method. It's an invention that makes inventing new knowledge vastly more effective and faster. With that knowledge, we've been figuring out new, better solutions to more and more problems at an accelerating pace. Now, back to the newly formed AGI, and why it heralds the way for a new epoch. During all this time, humans have completely lacked access to their own design. They have been unable to optimize how they think and how their brains work. Our brains work essentially the same as they did ten thousand years ago, on an architectural level. The fact that people now are more intelligent is actually evidence that giving the brain more resources does help it work better. But now we have a new creation, designed by the reasonably powerful optimizer, the human brain, and so at the start already more streamlined than evolution-built brains. And it has the ability to improve on itself. It can contain large amounts of knowledge, and it can work faster than a human brain can. It's located in the design-space for minds, at a point that is reasonably high, but willing to optimize itself further. It can search this space faster than any human, and more importantly it can change itself as it finds better solutions. And every time it finds an improvement, the speed of finding more improvements increases. Also, often enough one innovation opens the door for many more (such as figuring out a combustion engine and then using it to design all sorts of new machines), and some innovations work out to improve your metaknowledge, which can drastically, multiplicatively increase the speed of object-level innovation - figuring out a better way to form your hypotheses in order to hit upon successful ones faster, or developing a branch of mathematics that lets you solve physics problems faster. Or just knowledge-level improvements, like inventing a new, stronger theory of quantum physics that allows you to harness properties others don't even know about, for instance. And some innovations on the object-level allow access to vast resources. Imagine the AGI inventing cheap and efficient molecular nanofabrication.

If you want to see an example of the sheer power of improvements in the most meta level of problem-solving, the architecture of the mind, look at a chimpanzee and compare it to a human. We're extremely similar in genetic makeup. 95% of the chimpanzee's genes are the same as ours. But our brains are slightly different in design, and we rule the world.

Eliezer Yudkowsky has an interesting series of posts on the matter, with loads of text to read through. Here's an easy one to get into without spending too much time or effort for just a glimpse into the idea behind the argument. Following the whole sequence is a good idea if you're interested in the subject, in my opinion.

As for boredom - I don't see why I would spend 8 hours doing 'work' and 4 hours 'relaxing by entertaining myself' if I could redesign myself. I would simply spend 12 hours 'entertaining myself by doing work I enjoy doing'. It is not as if I could not simply choose what I want to like doing. Or, say, suppress the emotion of boredom until I'm done working as fast as I can to make myself powerful enough that I don't need to work. Or any of a whole number of other solutions that open up when you have freedom over your own mind.

[SiuiS]

"At a certain level of analysis"? It's context. At the level of individual elementary particles, there are no such things as pressure, temperature or an arrow of time. But as we get more and more particles these concepts become more and more useful. In the same way, an AI or a human could be deterministic machines at one level of analysis, but not at another. When we collect enough cells, we get complex patterns and phenomena such as free will become important. And I'm wondering if people are looking at AIs as deterministic machines and humans as free-willed subjects, and that is the reason for the stubborn refusal to date an AI.

But I also have a fever and one hell of a cold, so this may just be me rambling.

Ah, that's a good explanation. Thank you.

I believe that the level of analysis at which humans aren't machines isn't relevant, beyond acknowledging that people with a squick-based aversion or basic, visceral reaction are likely to view it as an important justification.

They shouldn't need to justify though. "Not for me ew" is cool.

That's true. I wouldn't date one.

Okay. That's what I thought you meant but I didn't want to assume and by the time I posted I forgot anyway. XD

[Starwulf]

I would argue that refusing to date a sapient AI on the principle that they are "not a person" or "only a machine" is, if not bigoted, at least uninformed. Personhood is not limited to humans (even on Earth); that's a very anthropocentric viewpoint. And humans are only machines. Refusal to date a sapient AI on the basis of lack of physical attraction towards their chassis or lack of interest in their personal qualities is, on the other hand, perfectly acceptable.

I wanted to stay away, because I'm obviously in the minority, but this kind of offends me. Calling a person who won't date a machine bigoted or misinformed is just wrong on so many levels I don't even know where to start. I'll leave it at that, if I go any farther I'm going to say something that gets me a warning. I do want to address one other thing though while I'm posting, aimed towards someone who quoted me earlier and said they found my viewpoint abhorrent/disgusting: I'm exactly the same, I find the viewpoints of those who want to humanize robots(no matter how sophisticated) and give them rights just as abhorrent and disgusting(the viewpoints, not the people!!!!!). They are MACHINES, parts, silicone, oil and circuitry. They are not human, they are not alive, they deserve no rights whatsoever. Just even trying to comprehend giving a robot rights honestly makes me physically ill.

[enderlord99]

I wanted to stay away, because I'm obviously in the minority, but this kind of offends me. Calling a person who won't date a machine bigoted or misinformed is just wrong on so many levels I don't even know where to start. I'll leave it at that, if I go any farther I'm going to say something that gets me a warning. I do want to address one other thing though while I'm posting, aimed towards someone who quoted me earlier and said they found my viewpoint abhorrent/disgusting: I'm exactly the same, I find the viewpoints of those who want to humanize robots(no matter how sophisticated) and give them rights just as abhorrent and disgusting(the viewpoints, not the people!!!!!). They are MACHINES, parts, silicone, oil and circuitry. They are not human, they are not alive, they deserve no rights whatsoever. Just even trying to comprehend giving a robot rights honestly makes me physically ill.

Have you ever read feet of clay by Terry Pratchett? Please do so, but keep a bowl with you, because you'll probably need to vomit a couple times. There's no metal, no electricity, no lubricating substances... but you'll probably need to vomit a couple times.

Also: robots shouldn't be designed in such a way that rights become an issue... but it's almost inevitable that it will happen anyway.

[Starwulf]

Have you ever read feet of clay by Terry Pratchett? Please do so, but keep a bowl with you, because you'll probably need to vomit a couple times. There's no metal, no electricity, no lubricating substances... but you'll probably need to vomit a couple times.

Also: robots shouldn't be designed in such a way that rights become an issue... but it's almost inevitable that it will happen anyway.

Just to lighten the mood a bit, the bolded part just really, reaaallly made me laugh ^^

[Lex-Kat]

I would. Artificial love is still better than no love.

[Bulldog Psion]

That depends:
Sexy fembot? Sure (as long as it's not trying to kill everyone or is obsessed with me that is)
Ipad? Nope.

Concise and to the point, sir.

[SiuiS]

I wanted to stay away, because I'm obviously in the minority, but this kind of offends me. Calling a person who won't date a machine bigoted or misinformed is just wrong on so many levels I don't even know where to start. I'll leave it at that, if I go any farther I'm going to say something that gets me a warning. I do want to address one other thing though while I'm posting, aimed towards someone who quoted me earlier and said they found my viewpoint abhorrent/disgusting: I'm exactly the same, I find the viewpoints of those who want to humanize robots(no matter how sophisticated) and give them rights just as abhorrent and disgusting(the viewpoints, not the people!!!!!). They are MACHINES, parts, silicone, oil and circuitry. They are not human, they are not alive, they deserve no rights whatsoever. Just even trying to comprehend giving a robot rights honestly makes me physically ill.

Humans named themselves after humanity not because humanity is what they have, but what they strive towards.

No one is saying they're human. They're saying that they are a person. Small difference, but it's less about if robots are humans and more an analogue of how we would view elves, dwarves, goblins, or even aliens. Once they're intelligent enough to have a society that can shape a planet, they get accorded the respect that humans themselves want.

[Eldan]

I'm not even expanding the definitions. I pulled out the textbook definition of "life" and explained how it applies equally well to organic and inorganic machines.

That really depends on your textbook, then. I have a few here that have "has cells" as a definition of life. Not that I necessarily agree with that.

[Eldan]

They are MACHINES, parts, silicone, oil and circuitry.

We are machines. Our parts are organs. Our silicone is proteins, our oil is, well, oil, our circuitry is small bundles of protein and fat. Our metal is a calcium matrix.

[Starwulf]

We are machines. Our parts are organs. Our silicone is proteins, our oil is, well, oil, our circuitry is small bundles of protein and fat. Our metal is a calcium matrix.

I fully disagree with that, on such a fundamental level that we will not be able to reconcile our disagreement, hence why I've not responded much during this thread.

[Eldan]

Can I still ask what you think humans are, then?

[enderlord99]

I fully disagree with that, on such a fundamental level that we will not be able to reconcile our disagreement, hence why I've not responded much during this thread.

{scrubbed}

[SiuiS]

Can I still ask what you think humans are, then?

Human, which machines are not.

Now, as it's been said to involve [Forum Banned Topics] twice now, with Starwulf specifically, we should probably not look to debate his stance. At the very best you will just be baiting him into getting himself in trouble.

[Eldan]

Not debate, necessarily. I'm just curious.

[Starwulf]

{scrubbed the original, scrub the quote}

{scrubbed}

Human, which machines are not.

Now, as it's been said to involve [Forum Banned Topics] twice now, with Starwulf specifically, we should probably not look to debate his stance. At the very best you will just be baiting him into getting himself in trouble.

I'm glad at least one person is willing to see that further debate on the subject of my opinion is not forum debatable. I will say this though, Just because I have a different opinion does not make me a bad person. Just because I'm not willing to treat a hypothetical sentient AI as a creature deserving of rights does not make me awful. Stop attacking my character people, because if that's all you can do, you are certainly no better then the awful construct you are making me out to be.

I am rapidly coming to understand what some others say when they talk about how unfriendly this forum really is to people who do not share popular opinion. It's kind of ridiculous that I'd be attacked just because I can not sympathize with a hypothetical construct of someones imagination. First I am told how abhorrent and horrible my opinion is by one poster, and then another attacks me accusing me of being some religious bigot. Got news for ya Enderlord, I"m not religious even a little bit, and talk of souls is wasted on me. Go back to the drawing board and find another way to attack my character.

[Eldan]

This forum does have a tendency do just sweep everything unpleasant under the carpet by deleting it, yeah. Doesn't mean it's not there.

[Teddy]

*snip*

Okay, let me explain my issue a bit more thoroughly. My problem lies with the assumption that the effort required to discover a better intelligence grows at worst linearly to the current level of intelligence. If this assumption holds true, then you'll get fast improvements indeed (and if it's sublinear, it'll even explode). However, practically no optimisation problem solutions scale linearly, meaning that every new iteration will require more time to calculate than the previous one, even with increased intelligence. Furthermore, everything about them is a tradeoff between time and memory complexity. If you can't afford all the memory required to keep yourself plus the entire big picture plus all current data in your optimisation calculations in mind at the same time (you basically won't, the memory complexity is generally terrible as well), you'll have to sacrifice some knowledge to free up memory, usually capping your awareness of the big picture and "old" or "distant" data, sacrificing some calculation efficiency as well.

Now, assume all improvements are baby steps compared to the time it takes to calculate them. With superlinear (and I'll dare to say at least polynomial) growth, you'll soon end up spending unfeasable amounts of time making only very slight improvements.

As for boredom - I don't see why I would spend 8 hours doing 'work' and 4 hours 'relaxing by entertaining myself' if I could redesign myself. I would simply spend 12 hours 'entertaining myself by doing work I enjoy doing'. It is not as if I could not simply choose what I want to like doing. Or, say, suppress the emotion of boredom until I'm done working as fast as I can to make myself powerful enough that I don't need to work. Or any of a whole number of other solutions that open up when you have freedom over your own mind.

Yes, this is exactly what I meant with "eliminating the ability to feel boredom" when I brought it up. Tailoring the conditions so that it never happens is pretty much the same thing as just ridding yourself of it entirely. I also think it would be just as prone to obsession (If you love making yourself more intelligent, why would you ever stop to do anything else?).

[Frozen_Feet]

They are not human, they are not alive, they deserve no rights whatsoever. Just even trying to comprehend giving a robot rights honestly makes me physically ill.

Let me offer a very different perspective then, as a factory worker who shares workspace with various sorts of automatons.

Computer programs of reasonable complexity require access to differing file sets of information. Without those files, a program can't do the job it was designed for. File access is, by now, a well-established concept in computer software and when you introduce any sort of AI to the picture (even a rudimentary, non-sentient one), granting access becomes synonymous to granting rights. Heck, several widely-used programs (Facebook, f. ex.) ask for rights to peruse your information, in those exact words.

In addition, any sort of a robot needs unobstructed space to do its job. While it'd be reasonable to say humans in abstract have right to stop a robot working and scrap it at any time, realize that in practice such rights are never granted to all humans regardless of who they happen to be. In practice, most people do not have permission to do anything to most complex machinery and will be sanctioned if they do.

In practice, this means automatons already have limited freedom of movement, freedom of knowledge and right to not be harmed. De jure we affix these rights to their owners, rather than the machines themselves, but de facto they are there, exactly like someone's duty to not do you harm conversely means you have right to live without being harmed.

It gets even more interesting when we take into account how legal systems already personify business companies and corporations, allowing them to own and transfer property or be sued as if they were singular persons rather than cluster of inviduals. An argument could be made that human cultures are AIs using human inviduals as logic nodes.

Those rights for machines and abstract entities are there because they help ensure a working society. It is a matter of pragmatism. Even in the case all computer AIs will forever be Philosophical Zombies, granting them the same rights and duties as real humans would prevent abuse, exploitation and unclear circumstances.

[Murska]

Okay, let me explain my issue a bit more thoroughly. My problem lies with the assumption that the effort required to discover a better intelligence grows at worst linearly to the current level of intelligence. If this assumption holds true, then you'll get fast improvements indeed (and if it's sublinear, it'll even explode). However, practically no optimisation problem solutions scale linearly, meaning that every new iteration will require more time to calculate than the previous one, even with increased intelligence. Furthermore, everything about them is a tradeoff between time and memory complexity. If you can't afford all the memory required to keep yourself plus the entire big picture plus all current data in your optimisation calculations in mind at the same time (you basically won't, the memory complexity is generally terrible as well), you'll have to sacrifice some knowledge to free up memory, usually capping your awareness of the big picture and "old" or "distant" data, sacrificing some calculation efficiency as well.

Now, assume all improvements are baby steps compared to the time it takes to calculate them. With superlinear (and I'll dare to say at least polynomial) growth, you'll soon end up spending unfeasable amounts of time making only very slight improvements.

Hm. Whereas, when I look at history of life, it seems to me that the more improvements are found, the faster new, even stronger ones tend to be found in turn. The complexity and depth of life seems to be growing at least close to exponentially, even now.

Yes, this is exactly what I meant with "eliminating the ability to feel boredom" when I brought it up. Tailoring the conditions so that it never happens is pretty much the same thing as just ridding yourself of it entirely. I also think it would be just as prone to obsession (If you love making yourself more intelligent, why would you ever stop to do anything else?).

Maybe I love something else, too? I love playing video games, but sometimes I also go out to play board games with friends, or watch a movie, or camp in the woods, or look at the stars. My utility function is pretty complicated, and I assume the same for any complex form of life.

[Teddy]

Hm. Whereas, when I look at history of life, it seems to me that the more improvements are found, the faster new, even stronger ones tend to be found in turn. The complexity and depth of life seems to be growing at least close to exponentially, even now.

Well, I'm not going out of my way to disagree with that (even though I might sometime in the future decide to argue the details ). I'm merely of the mind that complexity doesn't equal intelligence.

Maybe I love something else, too? I love playing video games, but sometimes I also go out to play board games with friends, or watch a movie, or camp in the woods, or look at the stars. My utility function is pretty complicated, and I assume the same for any complex form of life.

True, but I do believe boredom exists for a reason, since it's so deeply integrated into our minds, and I'm curious to how we would become without it...