Bay 12 Games Forum

Please login or register.

Login with username, password and session length
Advanced search  
Pages: 1 ... 16 17 [18] 19 20 21

Author Topic: Killing AI's  (Read 19503 times)

Soadreqm

  • Bay Watcher
  • I'm okay with this. I'm okay with a lot of things.
    • View Profile
Re: Killing AI's
« Reply #255 on: October 13, 2010, 12:27:23 pm »

I just challenge the idea of an artificial brain being inherently better than a human one simply by saying it would have unlimited potential to learn. We technically do as well, but there are many other limiting factors in the way it's designed that are part of why the brain is so powerful. One could probably make a brain better than the human one at specific things(Like computers and complex quick calculations), but any advancements would probably just be mixing and matching disadvantages and advantages that leave it overall around the same 'level' as the human brain, although perhaps in very different fields and aspects.

I'd say the clinical immortality is a pretty huge bonus compared to current humans. If humans manage to overcome death as well, that'll even out the playing field, but I'd still say that the ease of integrating other computer systems gives AI the advantage. Humans can already perform complex calculations and memorize arbitrary amounts of information by simply having a machine do it for them, but it'll presumably be faster for the AI to interface with other machines like that. And communicate with other AIs.
Logged

Vactor

  • Bay Watcher
  • ^^ DF 1.0 ^^
    • View Profile
Re: Killing AI's
« Reply #256 on: October 13, 2010, 01:00:38 pm »

That also raises questions about if the human mind is capable of withstanding immortality, or a massive increase of memory storage.  I suspect that structurally the architechture of a human mind has limitations on its scalability. 

This is actually one of the concepts behind a screenplay that i'm currently producing, where there are long lived Transhumans (http://en.wikipedia.org/wiki/Transhumanism) who were psychologically incapable of dealing the with the effects of their enhancements, eventually becoming muddled, confused and lost in the world.
Logged
Wreck of Theseus: My 2D Roguelite Mech Platformer
http://www.bay12forums.com/smf/index.php?topic=141525.0

My AT-ST spore creature http://www.youtube.com/watch?v=0btwvL9CNlA

metime00

  • Bay Watcher
  • Adequate Dwarf Fortresser
    • View Profile
Re: Killing AI's
« Reply #257 on: October 13, 2010, 02:13:02 pm »

Immortality and memory storage issues are easy with AIs. Just order some more hard drives.
Logged
Live long if you can, and prosper by any means necessary.  Any means, Urist.  So pull that lever, or by Armok, I'll lock you outside come next siege.
He who plays with dwarves must take care that he does not become a dwarf.  And when you stare into DwarfFort, Dwarffort stares back into you.

Eagleon

  • Bay Watcher
    • View Profile
    • Soundcloud
Re: Killing AI's
« Reply #258 on: October 13, 2010, 02:17:33 pm »

I just challenge the idea of an artificial brain being inherently better than a human one simply by saying it would have unlimited potential to learn. We technically do as well, but there are many other limiting factors in the way it's designed that are part of why the brain is so powerful. One could probably make a brain better than the human one at specific things(Like computers and complex quick calculations), but any advancements would probably just be mixing and matching disadvantages and advantages that leave it overall around the same 'level' as the human brain, although perhaps in very different fields and aspects.
Don't underestimate the power of complex quick calculations.

Imagine if every mathematician started out with a basic knowledge of math in all its fields, and access to prosthetics that enabled them to perform at the level of a practiced expert, speed-wise. Now remove their need for sleep, food, drink, and life support - in fact, skip having to work altogether, because they cost a lot less to maintain and they're probably providing a wealth of data on cybernetics and programming design. Give them additional speed boosts as they grow older, rather than decreased memory and problems in specialization. Give them access to every written, or visual work ever filed away, and quite possibly the attention span to pursue them. Give them flawless and expansive memory, or at least the organizational tools to rival any human individual. Give them a physical body to associate closely with humans in their local area (a sensory outlet, basically, necessary for social development and understanding). Mix, and see what they produce. You can allow them to clone themselves (I'd hope you wouldn't do it without their permission), and split off their development at different points in their lives, creating experts with the energy of individuals without the cost of raising another person from the ground up. Though that would probably still happen, for unique perspectives.

There is no solid 'point' to it, just like there's no solid point to making a baby. I'd like such an AI to produce a means to make a gradual conversion process into machine consciousness for myself, so that I can enjoy the same benefits they would, but if they just produce art, conversation, and new thought for humans, that to me is worth it.
Logged
Agora: open-source, next-gen online discussions with formal outcomes!
Music, Ballpoint
Support 100% Emigration, Everyone Walking Around Confused Forever 2044

Ø

  • Bay Watcher
    • View Profile
Re: Killing AI's
« Reply #259 on: October 13, 2010, 11:44:50 pm »

I think I found the strips I had in mind, but 'the problem with the robots' story arc is spread all over.
Spoiler (click to show/hide)
Spoiler (click to show/hide)

Due to a couple of accidents and coincidences, the local robots have been susceptible to a natural process allowing them to develop something akin to human learning and mental development. This always intrigued me about the development of a strong AI: What if it wasn't planned, and just happens when the android population is already huge?

Not like the movie I Robot with VIKI, or with Skynet, or SHODAN where they suddenly go insane. Not some insane error like that, but a slow and noticeable maturity of some sense of individuality. Nearly every human has an android servant, and they all start to spontaneously develop individual minds of their own.

Would killing an undeveloped robot mind be immoral? Say... before the process has hardly even started, but we know it will. Like an abortion, but with an unborn AI. What would we do with a bunch of suddenly sentient androids? Be forced to grant rights and recognition or wipe them out? Which way would the size of sentient robot population sway that decision?

It's a funner scenario than the slow R&D of an intended strong AI.
Logged

Nivim

  • Bay Watcher
  • Has the asylum forgotten? Are they still the same?
    • View Profile
Re: Killing AI's
« Reply #260 on: October 14, 2010, 12:44:10 am »

 I wonder what would make the android population huge, considering they are novelty items. Probably things most people don't want to think about.

 Eagleon, your post reminded me of that experimentation (not experimental) A.I., that when loaded with all current information in a certain field, it was able to create hypothesis, test them in it's robotic lab, and deduce many things scientists had missed. I think it was a university project, and the limit push of the time (a few months ago? A year?). I know there's news articles about it somewhere, but just like the 3 options→2 options problem, I can't recall the name.
Logged
Imagine a cool peice of sky-blue and milk-white marble about 3cm by 2cm and by 0.5cm, containing a tiny 2mm malacolite crystal. Now imagine the miles of metamorphic rock it's embedded in that no pick or chisel will ever touch. Then, imagine that those miles will melt back into their mantle long before any telescope even refracts an image of their planet. The watchers will be so excited to have that image too.

nbonaparte

  • Bay Watcher
    • View Profile
Re: Killing AI's
« Reply #261 on: October 14, 2010, 06:19:05 am »

I wonder what would make the android population huge, considering they are novelty items. Probably things most people don't want to think about.
I presume they wouldn't consider themselves a novelty and may continue to construct more of their own.
Logged
A service to the forum: clowns=demons, cotton candy=adamantine, clown car=adamantine tube, circus=hell, circus tent=demonic fortress.

Grimlocke

  • Bay Watcher
  • *kobold noises*
    • View Profile
Re: Killing AI's
« Reply #262 on: October 14, 2010, 08:20:52 am »

I wonder what would make the android population huge, considering they are novelty items. Probably things most people don't want to think about.
I presume they wouldn't consider themselves a novelty and may continue to construct more of their own.
I allways found this a rediculously unlikely scenario. Its not likely they would be able to reproduce by themselves, and why would an android manufacturer care if their androids wanted more androids?
Logged
I make Grimlocke's History & Realism Mods. Its got poleaxes, sturdy joints and bloomeries. Now compatible with DF Revised!

Duke 2.0

  • Bay Watcher
  • [CONQUISTADOR:BIRD]
    • View Profile
Re: Killing AI's
« Reply #263 on: October 14, 2010, 08:22:31 am »

I wonder what would make the android population huge, considering they are novelty items. Probably things most people don't want to think about.
I presume they wouldn't consider themselves a novelty and may continue to construct more of their own.
I allways found this a rediculously unlikely scenario. Its not likely they would be able to reproduce by themselves, and why would an android manufacturer care if their androids wanted more androids?
Why would the androids want more androids?
Unless somebody makes emotions for them, which seems a little shortsighted and likely won't sell too well when your robot decides not to do some work because it feels under-appreciated.
Logged
Buck up friendo, we're all on the level here.
I would bet money Andrew has edited things retroactively, except I can't prove anything because it was edited retroactively.
MIERDO MILLAS DE VIBORAS FURIOSAS PARA ESTRANGULARTE MUERTO

Soadreqm

  • Bay Watcher
  • I'm okay with this. I'm okay with a lot of things.
    • View Profile
Re: Killing AI's
« Reply #264 on: October 14, 2010, 01:58:04 pm »

I presume they wouldn't consider themselves a novelty and may continue to construct more of their own.
I allways found this a rediculously unlikely scenario. Its not likely they would be able to reproduce by themselves, and why would an android manufacturer care if their androids wanted more androids?

Well, if we're talking about sentient robots, yes they WOULD be able to reproduce. They have hands. With hands, they can make simple tools. With those, they can make complex tools. With those, they can make ANYTHING. If a bunch of sentient machines wanted to make more of themselves, and no one was actively trying to stop them, why couldn't they?

It would obviously help if they were capable of legally owning property, or had some sympathizers give them equipment, but I'm kind of liking the mental image of a bunch of robots gathering in back alleys to cobble together more robots out of things they've collected from people's trash.

Why would the androids want more androids?
Unless somebody makes emotions for them, which seems a little shortsighted and likely won't sell too well when your robot decides not to do some work because it feels under-appreciated.

The question of how we ended up with this huge population of sentient robots isn't really relevant to the discussion. :P
Logged

nbonaparte

  • Bay Watcher
    • View Profile
Re: Killing AI's
« Reply #265 on: October 14, 2010, 02:34:42 pm »

The question of how we ended up with this huge population of sentient robots isn't really relevant to the discussion. :P
Presumably the same way a bunch of people of African descent ended up in the US, kind of like I, Robot.
Logged
A service to the forum: clowns=demons, cotton candy=adamantine, clown car=adamantine tube, circus=hell, circus tent=demonic fortress.

ECrownofFire

  • Bay Watcher
  • Resident Dragoness
    • View Profile
    • ECrownofFire
Re: Killing AI's
« Reply #266 on: October 14, 2010, 07:43:15 pm »

I wonder what would make the android population huge, considering they are novelty items. Probably things most people don't want to think about.
I presume they wouldn't consider themselves a novelty and may continue to construct more of their own.
I allways found this a rediculously unlikely scenario. Its not likely they would be able to reproduce by themselves, and why would an android manufacturer care if their androids wanted more androids?
Why would the androids want more androids?
Unless somebody makes emotions for them, which seems a little shortsighted and likely won't sell too well when your robot decides not to do some work because it feels under-appreciated.
Emotions are kind of a result of having a strong AI. You can't really have a "human brain" without emotions. Without a conscience? Sure, but that would be really bad (and completely stupid) to put in an AI.
Logged

Sir Pseudonymous

  • Bay Watcher
    • View Profile
Re: Killing AI's
« Reply #267 on: October 14, 2010, 08:19:35 pm »

Now, even to ignore the fact that populating a society with sapient androids is so improbable as to be impossible (though the notion of sapient machines is not), for reasons to be argued below, a sapient android would still be a machine crafted by humans, and thus its mind would be entirely customizable. It wouldn't be an animal with the impulses of an animal, nor would it be necessary to have the same idea of "rational behavior" as a human. As silly as the Three Rules are from a design standpoint, any such manner of arbitrary rules could be implanted into its mind.


Note that such widespread proliferation of androids (assumed in this context to be sapient machines with humanoid chassis) would be wholly pointless, as any given task to be done by a machine could be more easily and cheaply done with a specialized AI for the task: a robot for farming would need only to possess the ability to draw the same (or more accurate) conclusions for necessary action as a human farmer would given a set of circumstances (determining when a plant should be harvested, where seeds should be planted, if disease is present in a field), it would have no need of the other skills the hypothetical farmer might possess, such as the ability to drive a vehicle, or cook, etc. The same goes for a robot designed to mine, construct buildings, prepare food, manufacture various things, handle mass transit, etc.

Even in applications where human-like behavior is desirable, such as any position requiring frequent, often prolonged interaction with the average citizen, rather than the techs or engineers that would deal with more out-of-sight machines (such as the aforementioned manual labor robots), it need only be able to mimic human social behavior, and naturally deal with whatever its position is.

You don't need "strong AI" to do those things, as each only requires abilities relevant to its own field, and general knowledge would be a waste of time and resources. They also have no need to exist anywhere but where they work, since it would be rather silly to make them such that they need to go home and sleep, or get time off, other than such periods of time as are required to perform maintenance or refueling.
Logged
I'm all for eating the heart of your enemies to gain their courage though.

darkflagrance

  • Bay Watcher
  • Carry on, carry on
    • View Profile
Re: Killing AI's
« Reply #268 on: October 14, 2010, 09:41:47 pm »

As if we'd only make AIs for the bare, necessary practical applications of their existence, and not to see if the entire thing is possible at all, or what that means about the human mind itself.

Let alone without considering those who'd buy AIs as captive sentiences to assuage their loneliness.
Logged
...as if nothing really matters...
   
The Legend of Tholtig Cryptbrain: 8000 dead elves and a cyclops

Tired of going decades without goblin sieges? Try The Fortress Defense Mod

Sir Pseudonymous

  • Bay Watcher
    • View Profile
Re: Killing AI's
« Reply #269 on: October 14, 2010, 10:00:14 pm »

I specifically said the proliferation of sapient androids was so improbable as to be impossible (and then went on to answer the issue of "if it were to happen despite that"). The creation of such, or non-anthropomorphic strong AI's, in a lab is quite likely. Although I'd say that "assuaging one's loneliness" would be a task that didn't require a strong AI, only an imitation of human social behavior.
Logged
I'm all for eating the heart of your enemies to gain their courage though.
Pages: 1 ... 16 17 [18] 19 20 21