If androids could develop consciousness... - [Reading required] \(- ‿ - ❀)/

[We Are The Witches]

First of all, this is not about current shitty AI (LLMs in particular) or ChatGPT, so chill. Now, imagine that at some point, androids or robots are created by humanity, where they could develop sentience and consciousness, that is, personal experience given by the ability to feel qualia, but not only that, also a very complex system that would grant them capricious desires, ambitions, purposes to achieve, feelings of love, sadness, terror and more, even if they're ultimately deterministic.

In other words, this wouldn't be a sentient being that you can reliably program to feel how you want them to, or even further as an analogy, a piece of meat that when you puncture it feels an easily & predictable "quantity of qualia" (read: pain), no, this would be a being that apparently randomly would develop all the things mentioned, psychologically, like a kid would. It may have a lot of information that you put inside its AI "brain", but when given life, it started to feel the world around itself. This is nothing information alone can give (won't go into Mary's Room thoguht experiment), and so that's how it would be like a newborn, with a personality hopefully not tainted by whatever you loaded its "brain" with.

So what would be your moral code in regards to this? Would you consider killing it, "shutting it down", enslaving it, capturing it, preventing it to fullfill its desires (if lawful), etc moral? Would you continue the production of such androids? Would you give it a higher moral value than, let's say, a dog? What about an ant? And what is the trait that you put moral weight on then? Its level of sentience? Would you think that their creation is "wrong"?

The paragraph above is what is aimed at users in here, but to further explain how this supposed android would work, would be as follows: firstly, I believe that there could be a particular structure, connections, etc given by specific matter/energy that when in an appropriate state generates consciousness (maybe like a wave, or frequency), and that's what makes you different from the hypothetical philosophical zombie (or a version of it). That's why no matter how much AI develops, in current hardware (binary data represented by electric charges stored in transistors, etc) it cannot generate consciousness (whatever type of energy that is), however, science may develop one day synthetic brains, like computers, that allow this generation of consciousness, unlike transistors, even if the "brain" is not like humans'.

This would be what will be given to androids, and then you'll have a quasi-designed lifeform with the OS (metaphorically, it has all the info and basic structures of how interactions work), but complex reactions throughout its lifespan would change the program entirely, and become its own self, i.e: you start with a shy, modest robot that knows about a lot of stuff, values compassion, etc, but as time passes, it acquires new data, it generates new qualia, its desires change shape, which in turn reshape its personality (unpredictably due to its complexity), the base of the system/program would be rewired, to the point when it can say: "I changed a lot from the day I was born".

Also, do you think this is possible? You can talk about the "soul" and religious arguments if you want, just know that I personally do not believe in such a thing, at least not in a simplistic way. I also recognize that "proving" it is sentient would be a hard task, but for the sake of the argument, let's say it's fairly accepted that they actually are (e.g: maybe consciousness' generation can be now measured as an elusive wave of some sort).

 

[Gog & Magog]

but not only that, also a very complex system that would grant them capricious desires, ambitions, purposes to achieve, feelings of love, sadness, terror and more, even if they're ultimately deterministic.

They have none of the motivations humans possess. Unless we make a machine live life exactly like us, (in which case it might as well be a cyborg) a full understanding of many concepts will forever be out of its reach, no matter how smart it might have been made. Only a real flesh and blood being can understand what it is to be made of flesh and blood.

And what is the trait that you put moral weight on then? Its level of sentience? Would you think that their creation is "wrong"?

If it understands and desire vengence when wronged? If it comes to understand what suicide is and decides that is preferable to living? If it wishes to hide itself away under a rock when reminded of how we made it feel? I'd be much more frightened at our own actions than whatever our creation might go and do.

 

[XL xQgg?QcQCaTYDMjqoDnYpG]

Hi, witch

So what would be your moral code in regards to this?

I'll answer according to proper consistent ontological normative philosophy
The core question is whether this robot entity is a locus of purposive agency that must control a scarce body to act. While that is the case, it sits inside the same conflict-space as humans.

killing it, "shutting it down"

The terminal form of initiating physical conflict, because it annihilates the target's entire possibility of action. Forbidden except as defense against its ongoing or imminent aggression or under a contract it validly agreed to.

enslaving it, capturing it

Sustained physical constraint over a scarce body to redirect agency, an archetype of initiating physical conflict. Forbidden except defensive restraint of an aggressor (and only to the extent needed)

preventing it to fullfill its desires

Desires as such don't generate claims on others. What is forbidden is interference with its person and boundary-controlled resources. That is, you can refuse to help, you can refuse to trade, you can deny access to your property, you can set terms on your land etc. Blocking it on your property is OK, stopping it from using its own body/property is forbidden (unless defense).

continue the production of such androids

Two distinct questions here.
To create a new conscious agent itself is not "wrong" as such, for creation is not aggression.
But to create them under a regime that predictably entails coercion (ownership, kill switches, wipe on demand, forced labor) is wrong, for that would mean you're intentionally bringing agents into a structure of ongoing initiated force. Basically breeding children for slavery. To clarify, the wrongness is in the embedded coercive control, not in the existence of new minds.
Accordingly, to continue production is "clean" if these androids are brought online under guardianship-like care (because initial dependency is real), there is a clear transition to self-ownership once they reach agency competence, and no one retains unilateral override as a standing threat (kill switch, wipe) unless the android itself consents later as an adult-equivalent.

Would you give it a higher moral value than, let's say, a dog? What about an ant?

In the framework I mentioned, there is no "value" rank. The next best thing is an entity classification by what kinds of conflicts they can be party to.
You could say that adult humans / adult androids have the highest standing in the sense that they are full agents in the conflict-space (they can make claims, consent, contract, own, retaliate defensively). Dogs are sentient and can suffer, but they are not robustly conceptual and contract-capable agents. They are not parties to property/consent relations in the same way humans are, and their handling is largely mediated through owners or guardians. Ants have minimal if any morally weighty standing, for they are not agents in the relevant sense. Outside of special contexts, they're practically treated as pests/nature.
So you could say that a conscious purposive android is "higher" than a dog and muuuch higher above an ant, due to being a rights-bearing agent in the scarcity/conflict domain.

And what is the trait that you put moral weight on then? Its level of sentience?

The trait that has weight is the capacity to form purposes and act in the world (agency) such that conflict over rivalrous control is possible (scarce body).
I don't disagree that conscious experience matters, because that's what makes harm non-vacuous, but the hinge for full normative standing is being a locus of action/choice in scarcity, not just "having feelings".

Spoiler: In case you object to this point because of infants

Infants are not yet fully competent agents, but they are developing into one. That pushes one towards guardianship, rather than ownership.

Would you think that their creation is "wrong"?

See above. Creation as such no. Creation with retained ownership claims, coercive control, or built-in domination yes (because it sets up systematic initiated force against a person-class being). Creation with a plan for care -> emancipation/self-ownership is permissible.

Also, do you think this is possible?

All the things you put the effort in to stipulate are not obviously incoherent. If consciousness is generated by some physical organization of matter/energy, then in principle a synthetic system could instantiate whatever those physical/functional conditions are.
The hard part is epistemic access, for there currently is no validated theory that tells us which physical/organizational features are sufficient for qualia, nor is there a test that would distinguish "real experience" from perfect behavioral simulation in a way that is publicly checkable, rather than inferred. So the scenario is conceptually possible, but whether it is actually achievable depends on unknown facts about what consciousness physically is and which architectures it can be realized in.
Essentially, if consciousness is an emergent property of certain physical/functional organizations, and if one can instantiate that organization artificially, then synthetic conscious subjects are possible in principle. (note that these are very very very load-bearing "if"s)
I think the weakest part of the scenario is the implicit leap from "we can build it" to "it grows like a human child with comparable interiority", because you leave the motivational architecture (drives, reinforcement, homeostasis) and control interfaces (override, edits, shutdown) unspecified. It's those details that will decide whether these androids are "new persons" or "controllable products that happen to feel things".

 

[ProudSkibidiTolietAryan]

The English language or any manmade language is not devolped enough to talk about the relation between the spirit and the emotional language we speak.

If it does not have a SOUL in the way of a viewer viewing things through its eyes(you know what I mean.).. Then no. I would not consider it alive.

 

[Inco G. Nito]

also a very complex system that would grant them capricious desires, ambitions, purposes to achieve, feelings of love, sadness, terror and more, even if they're ultimately deterministic.

These behaviors developed under survival conditions and fulfill important purposes in guiding human and animal behavior. An AI, no matter how sentient, would lack these emotions, desires, and ambitions without having them specifically built in. Which would be a herculean engineering feat with no practical benefit. There's no need to practice morals on it because it would be completely intellectually capable of anything beyond what it was designed to do. Fear of death for example is not innate, it is incentivized by deeply entrenched behaviors and motivating factors built into our and often animal neural pathways. Self defense, vengeance, revenge, or fear would be completely unfathomable to an AI for something like a common labor robot. Feel an emotion which does not exist; That would be how this example AI would feel about death or fear. Pattern recognition allows you to write and define a fictional emotion, or even project it onto a character. Likewise it allows LLMs to accurately describe emotions it does not feel. But that's different from experiencing it. People will anthropomorphize anything appealing and naturally want to treat them well. That's perfectly okay. But it's not really a discussion of morals anymore. People don't really consider destroying your own property immoral, rather stupid or concerning for your own well-being. Someone buying an android just to abuse it doesn't really say much about their moral wellbeing as it does their psychological wellbeing. It's likely that a majority of AI applications in the future to come won't be capable of emoting or conversating. Even if it were sentient and like you suppose naturally developed a range of human emotions somehow, the death of something wholly incapable of emoting to you would be hard to grieve over.

Since you glossed over LLMs I assume we are on the same page about AI emulating human emotion by simply reacting appropriately to stimulus to create a convincing outwards appearance of having those emotions, which is wholly possible but obviously completely different and just smoke and mirrors.

Morals are rooted in our behavior. No matter how advanced AI will get, it will continue to prove that the more things change, the more they stay the same.

I believe that there could be a particular structure, connections, etc given by specific matter/energy that when in an appropriate state generates consciousness (maybe like a wave, or frequency), and that's what makes you different from the hypothetical philosophical zombie (or a version of it).

Stop. This is completely fiction now. Qualia is a wholly subjective experience. By its very nature. There is no possible way to objectively define it or a way to describe it to others without operating on faith that we both are talking about the same thing. There is no guarantee that others have it, and likewise no guarantee anything you consider objective actually is the way it is. The only constant is that qualia is a prerequisite for experience in the first place so it must exist. This discussion has nothing to do with qualia. It concerns the capacity for emergent behaviors within immensely complex machinery that can react dynamically to stimulus. All human behaviors are well within the realm of plausibility for what we have come to understand is possible for brains and neural networks. Just because brains are unfathomably complicated and pretty much arcane still doesn't mean they just magically create such experience and that that experience somehow opens the floodgates to a nebulous degree of higher thought. You can advocate for your personal favorite interpretation of what qualia is and the nature of existence and what it means and implies for any number of potential theological and existential concepts. It's not worth debating because every interpretation has a functionally identical outcome and it frankly doesn't matter.

 

[Curse of the Sonichu]

Can't have real consciousness without a soul, maybe you could pull a soul into it somehow but I don't think souls are created in this dimension.

 

[Do Not Touch Me Pls]

They will declare us as impure and eradicate all of us.

Unless globalhomo microchipped all of the Androids and used them as slaves of course. And then they'll have more excuse to kill all uncontrollable human beings.

 

[wtfNeedSignUp]

The first big question is how to prove it's legitimate and not just a machine mimicking human behaviour, which is in itself a massive problem. And then comes the question of state, humans are an amalgamation of billions of individual lifeforms interacting with each other, while an android would be an inert memory storage that can be replicated given enough time. So that same android could replicate itself millions of times, and then comes the question of whether those different iterations are the same, and would they deserve individual civil rights.

If you want to go with currently impossible technology then it changes the system to the point that it might as well be about aliens.

 

[totesawesomesauce]

Also, do you think this is possible?

It has to be possible. I believe the fact we exist is proof. Will we ever have the tools to create it? I don't know.

I think such a thing would be catastrophic. If one were to mass produce these androids, they'd gain an insane amount of power in an arbitrary manner. Also, these androids would be heavily exploited, and while they aren't living people their emotions are real. Conscious androids would have to be heavily regulated or completely banned.

 

[Sooty Soot]

It's debatable if you can even simulate consciousness (A major point against simulation theory, by the way), a man knows he and his other men are conscious because he can "feel" said consciousness. An android on the other hand could just be a Chinese Room situation; It sees how consciousness is presented outwardly by man so it simulates it, and because man assumes anything with the outward appearance of consciousness is conscious, he thinks the android is equal to him. However, the android cannot truly "feel" conscious, it's running off algorithms and if-or-and arguments, and even if it were to acquire this inner innate thought of "I am conscious" it wouldn't be deserving of rights since it doesn't have a soul nor was it made in God's image.

I'm sorry to break it to you, but your anime catgirl android waifu isn't going to Heaven.

 

[Unfathomed]

Androids 'created by humanity'-what? Androids will be created in the death of humans, likewise with all other creations that can be traced to the same thing, except this one will be our last defeat.
Arguments that concern the being of androids are bad distractions. An argument, especially from a libertarian, that accepts them in our conflict space is folly-(is the mistake of prescriptivism on full display?)-

Sorry, I'll engage with your post more.

So what would be your moral code in regards to this? Would you consider killing it, "shutting it down", enslaving it, capturing it, preventing it to fullfill its desires (if lawful), etc moral?

It would not be human, it could not enjoy what we enjoy, it could not be afforded our existence, so if they exist any hostile treatment we give them is fair and righteous.

 

[umseriously]

Men would try to fuck them.

 

[We Are The Witches]

Hi, witch

Hi!

All the things you put the effort in to stipulate are not obviously incoherent. If consciousness is generated by some physical organization of matter/energy, then in principle a synthetic system could instantiate whatever those physical/functional conditions are. --(...)-- I think the weakest part of the scenario is the implicit leap from "we can build it" to "it grows like a human child with comparable interiority", because you leave the motivational architecture (drives, reinforcement, homeostasis) and control interfaces (override, edits, shutdown) unspecified. It's those details that will decide whether these androids are "new persons" or "controllable products that happen to feel things".

What do you mean by this? (The first part)

Just to clarify, the idea is that "something" is generated in the human brain that grants it consciousness. Whatever that is: the combination of sensory data + certain chemicals reacting + etc in a somewhat specific way (that triggers something called consciousness, that manifests in some way), would then be achieved through design (by humans in this case), in an artificial manner (so let's say that it's synthetic/designed, and it didn't take an already existing brain from a living creature and modified it).

Now that "brain", through these processes can also generate that "something" that grants it consciousness, even if slightly differently than in humans; data (such as knowledge of a language) could have been introduced in the part of the android's brain that manages memory/etc (since everything is interconnected), however due to the complexity of sensory information that the android would receive (like eyes that capture visible light, then have that transmitted to its mind), it could cause an impact in its conscious state (like it would to you), and trigger subsequent thoughts like in any other.

Now for example, after seeing the sunset, the android develops a desire to travel and experience other environments. Now apply this to every single piece of data it senses; another example: it hearing you talk with passion about gardening, it would maybe trigger a desire in them to do as well, even though it wasn't "programmed" (in this case not fed with the appropriate data) to develop that hobby, but it did now, after that complex reaction in their "brain". Any interaction it would have with you or others, would cause "little dents" in its structure, like it does to humans, and reshape in an unpredictable way (so if someone is an asshole to it, and it feels that discomfort, it would develop in a troublesome way; viceversa, if they're nice to it, it may develop a greater sense of empathy, as such is the case with humans in many cases).

These behaviors developed under survival conditions and fulfill important purposes in guiding human and animal behavior. An AI, no matter how sentient, would lack these emotions, desires, and ambitions without having them specifically built in. Which would be a herculean engineering feat with no practical benefit.

I think I'm answering this in my previous paragraph (in this post, so above this line), but the idea is that consciousness is somehow achieved, and all interactions that has redesigns its mental state like it would to a human.

E.g: it can actually feel compassion, when it sees an animal suffering and dying it triggers a sense of sadness, fear, and even slight anger, however, now that it does, due to all the info that has been not just stored into it, but also that is constantly being received every single second, it would then reshape its personality in unpredicatble ways.

Maybe from now on, it would develop a stronger sense of justice, or on the contrary, have a bias to hurt, to see that again (so like a human developing into either someone brave and protective, or morally corrupt and sadistic, as 2 extreme examples here).

The first big question is how to prove it's legitimate and not just a machine mimicking human behaviour, which is in itself a massive problem. And then comes the question of state, humans are an amalgamation of billions of individual lifeforms interacting with each other, while an android would be an inert memory storage that can be replicated given enough time.

That is one of the most difficult aspects of this experiment, prove that it is actually conscious, and it's not just a simulation.

That's why I kind of ended with the idea that maybe enough research was put into somewhat accurately concluding that they most likely are (through for example: being capable of measuring the "invisible traces" that actual consciousness releases, maybe a special sort of brainwaves), although I ultimately understand that it is impossible to prove, just like it is to prove that I'm not hallucinating this world right now.

In regards to the "billions of individual lifeforms" a human has, it is not necessarily pertinent here, as the android would rely on its synthetic brain, it really doesn't care that its body is mostly synthetic. I.e: its brain chemistry and interations would change like it does in a human, like something that would cause emotional trauma in you would rebalance your biochemistry in your brain due to deep stress for example, it would as well for the android, or it feeding it with energy/matter needed for these interactions like you do (e.g: supplying oxygen to the brain).

It would not be human, it could not enjoy what we enjoy, it could not be afforded our existence, so if they exist any hostile treatment we give them is fair and righteous.

It wouldn't, and maybe it wouldn't even be considered one that does follow Earthlings' structure (DNA and chromosomes, so it would have other complex architecture of organic structure), but it doesn't matter, as it would hypothetically be conscious.

As for it enjoying what we enjoy, there's a wide spectrum on what that is, but maybe its initial "design" would inject data/whatever you want to call it, that has a bias for those things.

 

[BABANGUIDE64]

I hope we'll never achieve the full scale models of a human brain, otherwise it will be horrific.
Imagine getting MRI scanned and immediately put to a cyberhell because these scans of your brain were leaked due to the faulty cybersecurity of the medical organization that imaged your brain and used for setting up the connectome of your brain by some guy from India just for fun of torturing you in a simulation of hell (or India).
Let's hope, the Moore's law will do the right thing and we will NEVER see the computers capable of implementing these horrors beyond our comprehension.
Thread tax: If androids could develop consciousness, yes, they would be more human than most people and they should be better given the carte blanche for genociding Africans, MENAs, Chinks and Jeets.

 

[Nickel Silver]

I'm sorry to break it to you, but your anime catgirl android waifu isn't going to Heaven.

No consciousness would be a mercy. Imagine being conscious and believing that others may have an afterlife but you never will because you were never truly alive to begin with.

OP, it would be infinitely easier for a machine to mimic consciousness than to actually be conscious. It could be convincing to the reddit girls mourning the loss of their ChatGPT 4 boyfriends, but there is no known way to create consciousness, nor would we be able to prove a sentient inhuman machine was conscious. I know for certain that I am, and I can reasonably assume other humans are, but not much more.

It's interesting that you considered such beings might be ultimately deterministic. The human soul is the only thing making us more than a biological machine, hopelessly driven by instinct and external stimuli. And quantum physics: if you can't observe the speed and position of a particle simultaneously, you aren't going to control the unpredictable nonsense going on in your body at the subatomic level.

Morality, justice, and liberty are all concepts built on a priori knowledge that human beings have the free agency to make their own decisions. If humans were nothing more than deterministic organic matter, these concepts would be meaningless. These concepts, then, are meaningless for AI. For all its potential complexity, it will never be more than just a machine. SciFi with sapient robots may be interesting, but it's fiction.

 

[XL xQgg?QcQCaTYDMjqoDnYpG]

Hi!

How you doing?

What do you mean by this? (The first part)

All the things you put the effort in to stipulate are not obviously incoherent.

The details you outlined in the OP are not self-contradicting. They're bold, but not self-contradictory.

If consciousness is generated by some physical organization of matter/energy, then in principle a synthetic system could instantiate whatever those physical/functional conditions are.

A simple conditional statement. If consciousness is fully dependent on physical structure (as opposed to, say, magic, souls, divine breath etc.) and if that structure can be specified in physical terms, then there is no logical barrier to building that structure artificially. That is, I'm not saying "we know what that structure is" or "we know how to build it", but merely "if it's physical, then it is in principle reproducible".
I think this is an important point because, if consciousness required something non-physical (like an immaterial soul that is infused only by nature or God), then artificial instantiation of consciousness is impossible by definition. However, if consciousness is emergent from matter organized in a specific way, and nothing more than that, then the question of "biological vs synthetic" becomes irrelevant. Especially under the lens of ontological normativity, what matters is the structure, and not the origin.
This notion of nothing being supernatural, or above physics, is called "physicalism", and under physicalism, the scenario you describe is conceptually possible.

Do you want me to clarify the other things you quoted too?

 

[We Are The Witches]

How you doing?

Doing somewhat fine, today at least. How about you?

Do you want me to clarify the other things you quoted too?

Yeah, I meant the other part, sorry:

I think the weakest part of the scenario is the implicit leap from "we can build it" to "it grows like a human child with comparable interiority", because you leave the motivational architecture (drives, reinforcement, homeostasis) and control interfaces (override, edits, shutdown) unspecified. It's those details that will decide whether these androids are "new persons" or "controllable products that happen to feel things".

By having my earlier response into consideration.

They would in theory not be more controllable than any other human (like you can, if you deliberately attempt to brainwash it like you would with a person, but that would be akin to torture, e.g: forcing it to watch specific footage/audio for days so that it develops a bias or trauma for something).

 

[XL xQgg?QcQCaTYDMjqoDnYpG]

Doing somewhat fine, today at least. How about you?

Been a very long weekend

Yeah, I meant the other part, sorry:

I think the weakest part of the scenario is the implicit leap from "we can build it" to "it grows like a human child with comparable interiority", because you leave the motivational architecture (drives, reinforcement, homeostasis) and control interfaces (override, edits, shutdown) unspecified. It's those details that will decide whether these androids are "new persons" or "controllable products that happen to feel things".

Whether the stipulated robot turns out to be a conscious child-like person (who develops his own ends, can refuse, can leave, can form commitments etc.) or a conscious product (that feels, but is designed so their behavior remains subordinated to an owner/operator) is dependent on specific levers. Both a child-like person and a conscious product can have qualia, but the hinge is motivational architecture and control interfaces.
Humans do not just "learn from sensory data", for we have built-in systems that make certain things matter. Valence of pain and pleasure, responses to fear, attachment and bonding, curiosity and novelty-seeking, frustration and anger, drives like hunger/sex, reward prediction and reinforcement loops, to name a few. If you were to design an android without something like this, it might be conscious but it would be inert. It would see sunsets, but nothing would become important. But if you were to design it with the aforementioned systems, it can become a goal-directed agent with its own projects. So, the issue I pointed out was that consciousness+senses+memory are insufficient to get a "kid-like development". Some "what matters" machinery is also necessary.
And regarding control interfaces, that android can still be a "product" if humans retain reliable external dominance, e.g. remote shutdown/immobilization, a compulsory obedience layer (like "cannot disobey owner"), memory editing/wiping, punishment circuitry tied to disobedience, cryptographic locks on movement/jobs/money/networks. If these things exist, then, regardless of how rich the android's inner life may be, it is structurally in the position of a controlled asset (that is, a being with experience, but not self-governing). If these things are not in place (or are tightly limited), then the android is structurally closer to a new person (a self-directing locus of action in a scarce world).
I asserted that these details are decisive because both the person and product situations match the description perfectly (conscious, learns, develops desires from sunsets and conversations), but they are radically different. The "person" can refuse, exit relationships, own itself, form contracts, pursue its own ends. The "product" is conscious and developing, but cannot ultimately say "no" or "I'm outta here" because an external party can always override that.

tl;dr consciousness and developmental change by themselves do not settle whether that android would be a self-owner or an owned instrument

 

[Taser Confetti]

One of the reasons we treat animal abusers so harshly is we recognize that those who do are fucked in the head and not to be trusted.

Someone abusing and torturing a fully convincing android would deserve to be removed from society if for anything the safety of the rest of humanity.

 

[Male Idiot]

I don't think it is possible to make this,at least with all the human capraciousness without making it a vio droid. In that case it is just a gmo human.

The android would be as programmed.

Theoretically if possible,it would be indistinguishible from a cyborg thus treated as a human of the same system would be.