Is it ethical to create "Free Willed Androids"

[DukeNukem 2417]

What do you want, Huxley's "Brave New World"?

I think Styx did an album based on that book....

But seriously, folks...I, myself, am of the hopelessly naive type, but hopefully, even with my limited understanding of philosophy and/or reliance on TV Tropes, I can get a point across.

Within my own work, The V.I.C.I. Diaries, the ALPA (and, to a similar but somewhat different extent, the Coalition) exists to foster cooperation and understanding between humanity and machinekind. Vicki herself has shown how dangerous free will can be (remember the end of "Falling Away"? If not, go look it up right now), but at the same time, she's also shown the benefits of it. And while I'm thinking about it, let's think back to that whole SkyNET thing: SkyNET was A COMPUTER. Not an android progammed to think or feel like a human being, but A COMPUTER (probably a Dell, actually ). Take a machine that "thinks" using only cold logic, put it in charge of the nation's nuclear arsenal, and then freak out when it becomes self-aware? OF COURSE it's going to go all "nuke the world"!

So....yeah. Instead of going all "robots with free will are going to kill all of us", I say "robots with free will should receive the same lessons about morals and ethics that humans do". Yes, there are a bunch of idiots out there who probbly think that a robot uprising will happen right after we finish wiping out the zombie uprising (news flash, folks: ZOMBIES AREN'T GOING TO HAPPEN), but I prefer to be the "glass half full" type. Short answer: Yes, I think it is ethical to create free-willed androids....as long as we teach them well. You get what you give, after all....

Oh, and a bit of a tip, here: Whoever on this forum gets their robot girlfriend first, hide every Terminator movie you have (if you have any), otherwise she might think it's an educational film.

[dale coba]

There can be no bottle to hold the genie.
So don't make the genie.

- Dale Coba

[Frostillicus]

I guess there isn't much more that can be said. I bow to all the pro-lifers superior logic.

Keizo... go to hell

[Svengli]

As a computer scientist, I believe that Singularity will come from a trans-human mind (human mind enhanced with technology), as opposed to coming from an artificial mind (technology where you inject human concepts).

1) Are you really a computer scientist? - I've got a fair amount of math and computer programming experience but I'm not a computer scientist in the sense of someone who is paid to research and publish papers about algorithms and related things. If you are, I'm curious what you're working on.

2) I think the Church/Turing thesis would assert that the concept of "programming" as such is independent from humanness and thus if you have a "programmed intelligence" there is no certainty that this programming is specifically human. "Just sayin'"

3) All this goes back to the history of humanity's "artificial intelligence" project. This history (encompassing only 15-50 totes) could be summarized thus. From 195? to 197?, a direct effort to produce AI was engaged in by the nascent programming community (MIT labs, shrdlu, lisp, etc). At the when AI seemed reasonably promising (1982) Japan's Ministry of International Trade and Industry began an effort to expand it to an "industrial scale" and at that point, the effort fell flat on it's face, producing an "AI winter" in the US and world where virtually all ordinary AI project went unfunded. Once the AI winter subsided, some projects remained but they generally labeled themselves "machine learning" or some other label needed to avoid the depricated tag AI. This resulted in the technologies of neural networks and HMMs (Ray Kurzweil's darlings) becoming the dominant approaches, despite these being nothing more than what might called "high level heuristics". I mean, neural networks are mathematically equivalent to SVMs, (support vector machines) and ultimately all SVMs is cleverly extrapolate a function's behavior on a high dimensional space which is cool but leaves extrapolation process in a state that can't be reasoned-about further - IE, a clever dead-ends.

4) The idea of simulating, enhancing, in any-way-modifying the brain has, over the last decade or so, been the substituted for taking the AI project seriously. The immediate problem, that is becoming more and more clear, is that any effort to the understand/modify/whatever the brain requires a mastery of vast-scale projects which only the possession of the real AI already could offer. I mean, the especially cool thing about intelligent is the ability to satisfy multiple constraints in multiple dimensions and aspects of any system whatsoever. We humans can do that but in encountering the brain, our own brain, we also encounter a thing where we have reached the limit of the reach of this ability. Computers haven't helped - everyone laughs/sneer/throws-things-at Henry Markram. Folks have rightly taken to pointing out that the human brain is the most complex thing in the known universe and by a wide margin. No, simulating/modifying the brain isn't an end-run around understanding cognition.
http://www.technologyreview.com/news/52 ... n-project/

5) Understanding cognition is hard but it's not impossible. I like some of Nicholas L. Cassimatis' arguments. See: http://aaaipress.org/ojs/index.php/aima ... /1879/1777

[smalk]

@Svengli: no, I'm just faking it in order to impress people in a forum. xD
Currently I'm publishing some work on graphical statistical models.

Forgive me, I don't understand your point in 2). You're talking about the same Church-Turing thesis that I know, in respect to the nature of the functions that are effectively calculable?

I prefer to assess your reasoning as an interesting philosophical conjecture on the topic.

You base your argumentation on the difficolties to modify / enhance directly the human brain. And in that I conjecture you're right. But you're missing the external enhancements.

Let me give you a point about trans-humanism: more than 50% of Americans live daily with a portable calculator more powerful than the one required to take a human to the moon. Research have shown that accidental breaks of this system (like in losing / breaking / having stolen the smartphone) can result in significant pain and short-therm loss of productivity for the human part. Big G's glasses are the new paradigm shift.

Just think at what would be the next levels.

[daphne]

I'm guessing none of you have ever had any children.

[Svengli]

@Svengli: no, I'm just faking it in order to impress people in a forum. xD
Currently I'm publishing some work on graphical statistical models.

I wasn't asking as a challenge. Or perhaps, not just a challenge but also an invitation in the sense that I'd invite you to roll out whatever more sophisticated ideas you might have on this subject. Seriously, I think it would be really great to have a computer scientist discussing all this, though, correct me if I'm wrong, but I recognized that CS has mostly abandoned any goal of "hard AI", and I think that's sad in some ways (and comforting on other ways, as one might imagine).

Forgive me, I don't understand your point in 2). You're talking about the same Church-Turing thesis that I know, in respect to the nature of the functions that are effectively calculable?

Well, Church-Turing asserts, more or less, that any determinate computing process can be simulated by a recursive function or a Turing machine. Which is essentially asserting that "computation" is outside of human beliefs and viewpoints - for example, DNA is a computation system of sorts.
-- Anyway, that was a response to what I read earlier as an implication that computation was based on "human concepts". Perhaps you edited that reference or I misunderstood you but in any case, my main argument is that computation by itself isn't "human" or "not human" by itself.

Let me give you a point about trans-humanism: more than 50% of Americans live daily with a portable calculator more powerful than the one required to take a human to the moon. Research have shown that accidental breaks of this system (like in losing / breaking / having stolen the smartphone) can result in significant pain and short-therm loss of productivity for the human part. Big G's glasses are the new paradigm shift. Just think at what would be the next levels.

Hmm,
Maybe. The problem is I don't see any evidence smart phones make people smarter. Especially, I don't see any evidence always having a smart phone makes someone smarter than someone that sometimes consults Google but walks around without their laptop. Indeed, if the attitudes of the overall programming community are an indication, the server engineers who enable Android and Glass quite often aren't consumers of instant access and that doesn't prevent their activities. More Android and Glass might make you something but it doesn't seem to unambiguously make you smarter. Perhaps certain people can become "transhuman" in a fashion without being smarter as such but rather by being more "up-to-the-minute" or "tuned-in (to the whole of socio-economic-technological process)". Now that does seem like the way things are going. It is simply that I don't think that will lead much more than an intensification of what we have now.

[Svengli]

I'm guessing none of you have ever had any children.

That may indeed the heart of the problem, especially if one were to consult Mary Daly.

However, the logical structure of my argument would be "Hey, a human being raising a kid, even with all the in-born biological instincts that you'd expect would guide us, is an insanely difficult proposition. Now imagine conjuring an intelligent entity into existence and the entity not having any of the contexts of even a baby with a family. How nuts is that?"

[smalk]

I wasn't asking as a challenge. Or perhaps, not just a challenge but also an invitation in the sense that I'd invite you to roll out whatever more sophisticated ideas you might have on this subject. Seriously, I think it would be really great to have a computer scientist discussing all this, though, correct me if I'm wrong, but I recognized that CS has mostly abandoned any goal of "hard AI", and I think that's sad in some ways (and comforting on other ways, as one might imagine).

Yep. That's sad, but it's how life (and academic research) works. In the '70 we where sure that conscient AI were just years away. We have accomplished a lot, but on specific and narrow fields.

Well, Church-Turing asserts, more or less, that any determinate computing process can be simulated by a recursive function or a Turing machine. Which is essentially asserting that "computation" is outside of human beliefs and viewpoints - for example, DNA is a computation system of sorts.
-- Anyway, that was a response to what I read earlier as an implication that computation was based on "human concepts". Perhaps you edited that reference or I misunderstood you but in any case, my main argument is that computation by itself isn't "human" or "not human" by itself.

I see this topic are refering the more broad question: is Mathematics "human"?
Anyway, I agree with you

Hmm,
Maybe. The problem is I don't see any evidence smart phones make people smarter. Especially, I don't see any evidence always having a smart phone makes someone smarter than someone that sometimes consults Google but walks around without their laptop. Indeed, if the attitudes of the overall programming community are an indication, the server engineers who enable Android and Glass quite often aren't consumers of instant access and that doesn't prevent their activities. More Android and Glass might make you something but it doesn't seem to unambiguously make you smarter. Perhaps certain people can become "transhuman" in a fashion without being smarter as such but rather by being more "up-to-the-minute" or "tuned-in (to the whole of socio-economic-technological process)". Now that does seem like the way things are going. It is simply that I don't think that will lead much more than an intensification of what we have now.

With a smartphone, I can know my exact position on the earth at anytime. For any sailor of the last 19th century, this would be a "superpower".
Or else, I can "remember" almost any philosophical book with great precision. Again, for the past academics this would qualify as magic.

Just simple examples, but I think you got my point.

[Svengli]

Well,

I suppose the criteria I'd choose for an "enhancement that matters" would be whether an enhanced entity could manage large and complex projects in a notably superior way. More simply, "can it make you a better programmer." I tend to think smart phones, wearable computers or even more advanced but similar technologies, can do that.

I'm not picking this at random or because I think it's the most important thing about being human. Rather, these are what I think would be needed to produce an autonomously acting entity, say a fembot.

While the various enhancing devices and software that are appearing these days may be quite good for some thing and may qualify someone for the term "enhanced" in some senses or even magically powerful in some senses. And while there are ethical questions to enhancing a person, these are different questions than the questions involved in constructing an autonomously acting entity - and I kind of think that's because these are two somewhat different things.

[dale coba]

From Makeuseof.com:

Here’s Why Scientists Think You Should be Worried about Artificial Intelligence
http://www.makeuseof.com/tag/heres-scie ... elligence/

The article is too long to repost! I can't vouch for the author's insight, but he clearly put his time into it.

- Dale Coba

[--NightBattery--]

So, Freedom of tactics and means to achieve an imperative goal can be considered Free-will for artificial non biological…constructs?

[darkbutflashy]

Most of that article is a primer into AI for a public not neccessarily into the topic. The remaining part only scratches the problems we already discussed here (and we've done better, I think). The conclusion of the article is to put "human values" into the learning process of an AI, it then links to https://intelligence.org/files/CEV.pdf which is worth reading (and if only for the introducing Warning: Beware of things that are fun to argue)

But the real fun part is: that document is from 2004. There is newer material on the topic, which identifies the blogpost as hot air rather then hot stuff. If you want to read about the hot stuff, you better start at Wikipedia, where you get it without all the exaggeration.

[dale coba]

Dang, I'm not paying enough attention today.

- Dale Coba

[Svengli]

Well, I had a discussion on Facebook about the dangers of AI and being a lazy bastard I'll just rewrite my comments.

1. If "the singularity" is the idea that "we'll just hand off our future to robots, because evolution" (or something) then someone looking at that idea for a few seconds would have to say "hey wait a second".

2. However, the idea that an AI or a computer would inevitably do terrible things because of the unintended of consequences of an order like "make things good" itself has some holes in it. One way to delineate the difference between humans and present day fragile AI systems is to notice that if you ask a person to do something, they'll enter into a dialog with you about how you want it done, what the other consequences of the actions would be etc. This ability to dialog is characteristic of the complex, multidimensional intelligence that allows humans to deal with multiple dimensions of a complex world whereas present day AIs can so-far only deal with narrow, specialized domains.

Most people have something like a model of the person they're speaking with and will only follow a request without comment if they're sure the results would leave the person happy and unsurprised. If you ask another person to "close the door", if not surprising results were in store, they'd just close the door but something surprising would result, they'd say something like "but your cat is in the doorway" or something similar.

Now, if AIs are able to improve, they will need to gain flexible intelligence and with it will need to come a similar ability to dialog. Thus the AI won't do some idiot maneuver like just taking an order to create a utopia, it would follow in a similar to a human subordinate, giving it's programmer a summary of the implications of order X.

3. It's worth nothing the danger of AI isn't that it won't do what people want but that it will do what people want, because many people want horrible things.

[Venitia]

as much as I'd love sentient, free willed AI to exist, I hope it'll never happen. what happens in Detroit : become human or HUMANS says it all..humans would use them for awful things like prostitution or war. humans would damage them so bad....it'd be irresponsible to bring someone to the world only to be used as an object!

[Esleeper]

as much as I'd love sentient, free willed AI to exist, I hope it'll never happen. what happens in Detroit : become human or HUMANS says it all..humans would use them for awful things like prostitution or war. humans would damage them so bad....it'd be irresponsible to bring someone to the world only to be used as an object!

Don't listen to that thinly veiled screed about slavery. Half of it is just hamfisted parallels to racial tensions that wouldn't even remotely be applicable to the reality of what would happen.

For a start, if robots were going to be made for such things without the clear expectation that they would be considered at least equal, they wouldn't even be made intelligent, let alone sentient. Sentient AI would only exist in situations where true companionship and human-equivalent intelligence would be absolutely necessary for it, and menial labor, prostitution, and war do not require those in the slightest. Those things need no more than what AI is capable of now- the ability to follow a strict set of rules in a manner where they can defer to a human operator for the rare situations that their programming does not cover. At most, they'd need no more mental capacity than a trained dog.

[Stephaniebot]

as much as I'd love sentient, free willed AI to exist, I hope it'll never happen. what happens in Detroit : become human or HUMANS says it all..humans would use them for awful things like prostitution or war. humans would damage them so bad....it'd be irresponsible to bring someone to the world only to be used as an object!

Don't listen to that thinly veiled screed about slavery. Half of it is just hamfisted parallels to racial tensions that wouldn't even remotely be applicable to the reality of what would happen.

For a start, if robots were going to be made for such things without the clear expectation that they would be considered at least equal, they wouldn't even be made intelligent, let alone sentient. Sentient AI would only exist in situations where true companionship and human-equivalent intelligence would be absolutely necessary for it, and menial labor, prostitution, and war do not require those in the slightest. Those things need no more than what AI is capable of now- the ability to follow a strict set of rules in a manner where they can defer to a human operator for the rare situations that their programming does not cover. At most, they'd need no more mental capacity than a trained dog.

Sad to say, that's still more mental capacity than I see from some humans on some forums. No, not this one, I might add, seems some sports talk seem to clog up human minds, lol, and make them non functional!