Sliding Scale of Robot Intelligence: Difference between revisions
Content added Content deleted
m (update links) |
m (update links) |
||
Line 2:
In fiction, it's [[Instant AI, Just Add Water|surprisingly easy to create AIs]], and their [[AI Is a Crapshoot|resulting morality is disturbingly random]]. However, you can depend on all AIs, Robots, and Androids presented in fiction having a range of levels of intelligence between [[Artificial Stupidity|Brick]], [[Do Androids Dream?|Human]], or [[Deus Est Machina|God]].
These are the five typical levels, though machine intelligences between these five grades are common:
# '''Brick:''' somewhat autonomous, stupid, uncreative robots. Your basic auto factory assembly line.
# '''Robo-Monkeys:''' Cute and intuitive, act just like animals. That is to say, clever and surprisingly instinctual.
# '''Average Joe Android:''' Very good memories and math skills, but usually lack interpersonal skills, [[Creative Sterility|creativity]] and/or [[Tin Man|emotion.]]
# '''Nobel-Bot:''' Just like Joe Android, but with a superior intellect capable of cracking most scientific problems in picoseconds.
# '''[[Deus Est Machina]]:''' The god AIs will be so far advanced that scaling them would be futile, but they usually have grades between each other, and may even still puzzle at [[What Is This Thing You Call Love?]].
These plateaus exist to help authors and readers wrap their minds around AIs. Human intelligence are used because writers and viewers can understand them more easily. Sure, they're [[Pick Your Human Half|logical/mechanical, but still basically human]]. God level intelligences are harder to write for, but that level of indecipherability can actually be easy to pull off with enough vagueness. The un-sexiest level is Brick, but you'll still get a few stories with massive armies of mindless automata which get put out of commission ''precisely'' because they're mindless. If they're autonomous and dumb, then it's likely being used to highlight the lack of malice inherent in a [[Nanomachines|Grey Goo]] or [[Zombie Apocalypse]] like [[Robot War]].
Some newer works might be all over the scale by saying that multiple dumb machines "networked" into a machine god (or at the very least smarter than human) intelligence capable of dissent.
Line 20:
* In ''[[Mahou Sensei Negima]]!'', [[Robot Girl|Chachamaru]] is a more or less average human-intelligence type robot with notable hacking skills, memory, and computing speed. Her master Evangeline has other [[Robot Maid|robotic servants]], most of whom seem to fall between brick and human levels; they appear to have a certain level of self awareness, but not nearly to Chachamaru's degree.
** It's mentioned every once in a while that this is because her "sisters" are pure robots, while she is a science/magic hybrid. Not to mention that her purely magic counterpart, the [[Creepy Doll|animated puppet]] Chachazero, can't even function if there isn't enough magic.
* Subverted in ''[[Ghost in the Shell]]: Stand Alone Complex''. The [[Ridiculously-Human Robots]] that appear are all non-sentient, and [[Logic Bomb|easily befuddled]] by the spider-like Tachikoma [[Incredibly Lame Pun|think-tanks]]. The Tachikoma discuss this, noting that humans would be intimidated by androids with human intelligence but are much more accepting of non-humanoid adorable robots like themselves having sentience.
** And in the original manga and movie there is the Codename 2501, a.k.a. Puppetmaster, who started out as a Brick with capability to learn, and became what is described in the final volume of the manga as an information god. In the aforementioned final volume, ''Man-Machine Interface'', the semi-AI descendants of the Puppetmaster-Major-fusion briefly plan turning every human with cybernetic implants on Earth into offshoots of themselves, but instead opt to create even higher forms of artificial consciousness. It's implied at the end that the age of machine gods is coming fast.
* ''[[Magical Girl Lyrical Nanoha]]'' uses varying levels of robot intelligence. [[Mecha-Mooks|Gadget Drones]] are of Brick-level intelligence and are mowed down by the dozens. Storage Devices also have Brick-level intelligence. [[Empathic Weapon|Intelligent and Armed Devices]] are somewhere between Brick and Human, capable of creating their own opinions and having conversations with their users but not to the level of humans. Finally, there's the Wolkenritter and Unison Devices that are Human level and [[What Measure Is a Non-Human?|pretty much considered as fellow humans by other characters]]. In one of the [[Opening Narration|Opening Narrations]], Vita, one of the Wolkenritter, idly wonders if she and the other Wolkenritter were nothing more than weapons like the Brick-level [[Kill Sat|Cradle]] they're facing before they met Hayate and gained human-like emotions and personalities.
Line 37:
* Averted by {{spoiler|Victor Mancha}} of ''[[Runaways]]'', who seems to be slightly more intelligent than your average teenager.
** This is something that [[Just Bugs Me]]: he is a cyborg (unless he's more of a [[Terminator]]).
* The italian sci-fi comic Nathan Never has mainly "brick" to "average joe" robots, some of latters with pretty human aspirations - desires; like having a girlfriend or going to the pub with some friends. A focal point of the stories where robots are protagonists is that, for [[A Is]] to truly grow "human", they have to be set free - in some way - from the "Three Laws".
** The series uses also a very interesting take at the "ridicolously human robot" concept: the first generation of autonomous androids was built without the "three laws" - relying on a distinct set of security limiters - and ended having almost completely "human" minds, so humans that they started strikes in order to obtain paychecks, holydays and respect for their rights. All the owners, then, sued the company that made them, that as a consequency went bankrupt.
*** Later, one of them tried to hide the "human nature" of he and his brethren in order to not be treated as a human... and forced to pay taxes.
Line 50:
**** Only the latter is relevant. [[Zombie Apocalypse|Zombies]] and [[Aliens]] have shown that inhuman and largely characterless antagonists can be quite interesting, though they are seldom ''amusing''.
*** It's correct in the sense that each squad of droids had a "command droid", which in turn was controlled by the central computer. It's also possible that CIS droids were autonomous, since there's no mention of any centralised control units, usage of those was based purely on Neimodian paranoia about control. It doesn't answer the question of why would they need to communicate with audible words at all, though.
*** The command droids, at least, need to interact with organic superiors. Perhaps the rank-and-file droids speak because the Neimoidians, noted for their paranoia, wanted to make absolutely sure that their footsoldiers couldn't plot against them (or even just mock them) behind their backs.
* A substantial amount of ''<nowiki>~2001: A Space Odyssey~</nowiki>'' is spent in discussions over the intelligence and emotional capacity of the H.A.L. 9000 computer that [[Master Computer|runs the spaceship]] USS ''Discovery''. It's generally agreed that HAL is of human-level intelligence, but while he has vastly superior powers of calculation (obviously), his emotional capacity and intellectual maturity are those of a child. This factors heavily into the explanation of the [[Logic Bomb]] that causes him to [[AI Is a Crapshoot|turn on the crew]].
Line 60:
== Literature ==
* [[Alastair Reynolds]]' ''Revelation Space'' series has primarily "Brick" style robots for the human designed servitor robots, and the [[Brain Uploading]] humans are effectively identical to their flesh counterparts. Reynolds' other novels however, play with it quite a lot. ''House of Suns'' has the Nobel-bot Machine People, and the godlike Vigilance.
* Averted in [[Iain Banks|Iain M. Banks']] ''[[The Culture|Culture]]'' novels: Sentient [[AI|AIs]] vary from humanlike to godlike intelligence (the Culture Minds), but there are lots of nonsentient ones as well. And there are some that are sentient, but somewhat below the human level, like the self-aware but simple-minded Culture shuttle that makes a brief appearance in ''Consider Phlebas.'' It was advised by more intelligent AIs to ignore the immoral behaviour of some nearby humans because its mind was human-like enough that it would have been shocked, but too simple for it to have been able to deal with the experience. Culture law specifies that everything over a given level of technological complexity must have an AI. This greatly annoys a character in ''Excession'', who deliberately chooses the most stupid AI he can get to control his second-skin environment suit. This is more about a) providing a nice user interface to complex but useful devices and b) providing intelligent safety systems to dangerous devices than making the lives of humans more difficult. An exceedingly powerful weapon developed by the Culture before such a decree was made features in one story in ''The State of the Art''. Though genetically locked to only be used by Culture-descended humans, one such being is {{spoiler|blackmailed into committing an atrocity with the weapon}}. The being in question observed how weak the weapon is compared to modern day devices, but as those weapons would be too smart to allow themselves to be abused this item is potentially devastating against less developed civilisations.
* The mobiles from the ''[[Young Wizards]]'' series might be an example: when they talk to each other about entropy the discussion goes right over the head of [[Teen Genius|Pre-Teen Genius]] Dairine. However, this might not be due to being smarter than Dairine, but because they each have a [[Great Big Book of Everything]] built into their memory.
* Averted in Dan Simmons' ''Hyperion'' where the robots are considerably more intelligent than humans but none of them have godlike minds.
* Artificial Intelligence Personalities in Donna Andrews' ''Turing Hopper'' mysteries have the capacity to upgrade themselves and some eventually achieve true self-awareness. Sentient AIPs mostly seem at the Human level (with greater-than-human expertise in the areas they were programmed to specialize in), but think faster and are better at multitasking and dealing with large quantities of information.
* ''[[The Hitchhiker's Guide to the Galaxy|The Hitchhikers Guide to The Galaxy]]'': "Here I am, brain the size of a planet and they ask me to take you down to the bridge." Implied to be a(n incredibly lazy/jaded/depressed/unmotivated) [[Deus Est Machina]].
** The doors of the ''Heart of Gold'': "[Satisfied hum] Glad to be of service!"
Line 87:
== Live Action TV ==
* Averted in ''[[Knight Rider]]''. KITT is smarter than any human, but not indecipherably intelligent. Most of his unrealism comes from being [[Ridiculously Human Robot|ridiculously human.]]
* ''[[Battlestar Galactica]]'': Skinjobs are as smart as humans, if not smarter. Raiders are trained pack animals, and Centurions are on the cusp of sentience.
** It's indicated that the Centurions ''were'' completely sentient, but the skinjobs went and lobotomized them as an ironic echo of the original robot rebellion. When the restrictions are removed by skinjob Cylon activists, they along with the Raiders return to sentience, and are not very happy at all.
** From [[The Movie]]: "His coat is burgundy. This is teal." Some of the skinjobs are [[Too Dumb to Live]].
Line 133:
** Well, it may think it is Godlike, but it's more like Human level -- sure, synthesized life, but humans by the year the game takes place could probably do that left and right. And sure, it altered reality, but only by using human-created technology. Finally, it needs you to run around doing everything for it since it is just a computer program after all.
*** Humans in ''System Shock 2'' don't seem to have much more sophisticated biological technologies than we do today. One part of SHODAN's mind constructed the ancestors of [[Hive Mind|the Many]] as a quick experiment. Humans may have created the FTL drive, but it took SHODAN to turn it into a universe-consuming reality warping weapon that would have made her a literal god. SHODAN needed the PC to begin the process as no other systems were handy and time was short. You are disposed of quickly enough once you usefulness ends and the AI becomes independent.
* ''[[Portal (series)|Portal]]'' has GlaDOS, a rather curious and quite insane AI which appears to be about the Nobel-Bot level.
** [[Portal 2]] gave us Wheatley, who's at the low end of "Average Joe Android" and was supposedly designed to be a moron in one of many attempts to make GLaDOS behave.
*** Aw, heck, a lot of the machines in Portal 2 (including the Enrichment Center itself) either belong or are treated as though they belong somewhere on the scale.
**** And there really isn't any way to tell where the Companion Cube is ( [[G La DOS]] does claim that it is sentient, after all).
* ''[[Command
* ''[[Final Fantasy XIII]]'' interestingly gives us all five in varying degrees. fal'Cie are literally God Machines capable of complex thought and philosophizing; then you have machines like the pulsework knights that only serve a single function. We see a group of hulking Juggernaut robots late in the game that normally operate like 1s but turn into 3s in order to protect one of the party members.
* [[Chrono Trigger]] and [[Chrono Cross]] together have robots which may cover the entire scale (except perhaps type 2): security drones and mannequins, average human-likes (Robo and his batch), Mother Brain, and then FATE.
Line 180:
== Real Life ==
* There's a very real question in computer science in the feasibility of "strong AI" versus "weak AI". A strong AI is the conscious, thinking, possibly self-aware entity. A weak AI is a glorified difference engine with language synthesis. Weak AI is a foregone conclusion -- the complexity of grammar and the limits of voice recognition are the only remaining stopgaps, and computers' ability to deal with those [[Science Marches On|is improving all the time.]] Strong AI is still a question of speculation, philosophy, life, the universe, and everything.
** "The only remaining stopgaps" indeed. Fluent natural language processing is an AI-Hard problem (from the mathematical term, NP-Hard). You probably can't do it without a true AI, but you can't make a true AI until you have fluent language processing...
** There is much more to a Strong AI than just communication capability, but that particular matter is hardly an absolute stopgap. A learning, language-specific weak AI should be perfectly sufficient once we get the algorithms right; research into human language-learning and related neurology will most likely play a major part in this development.
* Present artificial intelligences are of course, on the Brick level with some in-roads to the Robo-Monkey level. Entities like Roombas, ASIMO, the Mars Exploration Rovers, Deep Blue and [[Jeopardy!|Watson]] are, while neat feats of engineering, all Bricks.
| |||