Instant A.I., Just Add Water
"Good news, everyone! I've taught the toaster to feel love!"
Judging by television, it would seem that any sufficiently complex computer inevitably becomes sentient. It just happens, automagically, while the builder's back is turned.
It doesn't matter that we do not yet have a thorough enough understanding of all of the mind's mechanisms to artificially duplicate the thinking process. Just add a handful of memory chips, a bolt of lightning
or some of that Imported Alien Phlebotinum
and Bingo! It's wakey-wakey for the BFC-2000.
A bit more modern take on the trope might involve instructions that unintentionally result in sentience, such as ordering a program to continuously adapt itself. Such adaptations may occur overnight. Then again, they may be in a situation that forces them to Grow Beyond Their Programming
. Additional shortcuts may include the assimilation of large bodies of information ("the internet" tends to be popular these days).
Given that such a being is not man-made, the use of artificial is probably incorrect
, but most people won't care.
The end result of this may vary. Said AI may turn against its creator
, or become a Friend to All Living Things
. It may develop a human-like personality
or remain as cold and emotionless as it was before. What becomes of an intelligence after it becomes... well... intelligent... is not covered by this trope.
This is, perhaps surprisingly, one theory amongst real-world researchers in artificial intelligence. Some believe that a necessary prerequisite for machine intelligence is a certain minimum complexity of the system that runs the software (i.e. keep throwing more chips in it until it gets smart). There are a few theorists who think that there is a possibility that, given enough
complexity, some form of intelligence just might spontaneously develop (this is, of course, an extremely simplified explanation of a vast amount of research in machine intelligence, but still relatively accurate). Companies like IBM are spending buckets of money on pure R&D to develop supercomputers with massive numbers of connections just to test these theories, which makes this closer to Truth in Television
than you might expect.
See also A.I. Is a Crapshoot
and Sliding Scale of Robot Intelligence
. Of course, you don't want to actually add water (that would be stupid
), because there's No Water Proofing In The Future
. Subtrope of Creating Life Is Unforeseen
open/close all folders
Anime & Manga
- Harumi of Irresponsible Captain Tylor.
- The first episode involves the testing computer falling in love with him.
- The Tachikomas in Ghost in the Shell: Stand Alone Complex develop intelligence (and whimsical, childlike personalities) after Batou replaces their synthetic oil with the organic variety. So, more of an Instant AI Just Add Oil. Of course, the fact that Batou also cares deeply about the Tachikomas (as well as most other machines) probably plays a larger role in this...
- It is probably more important that he only gives organic oil to "his" Tachikoma. This favoritism starts giving them a sense of their own individuality. They become "self-aware".
- The Major's stated message from the end of the series about the Tachikomas is that curiosity was what drove them to individuality.
- Amusingly however, except for the AI of the Puppetmaster and the Tachikomas, digital AI with true personalities (and Ghosts) are notably few. Apparently, they are rare things indeed, despite a constant media paranoia about them.
- They were already programmed with their childlike personality and curiosity, as their AIs were built with adaptive programming. All the oil situation did was cause one to adapt to being an individual because Batou treated it like one. This is all made explicit in the second season when they arrest the Tachikoma's designer. Oh, and as I understand it, all the iterations of the series take place at a point when AIs are beginning to emerge.
- Morganna Mode Gone in .hack. It's not clear whether her creator intended her to be sentient, but she became so anyway, and immediately began screwing things up. A surprisingly large number of AIs unimportant to the story begin popping up in the game the series is centered on as well, though this is probably to be expected, since The World was secretly programmed to be an AI birthplace.
- Morganna's problem was being unable to do anything constructive with her sentience. She became locked into her purpose as stated and could see nothing else when she tried to think outside of that box. She procrastinated the birth of Aura for so long, then repeatedly damaged herself by breaking off to form the Phases, that eventually she became unable to rationalize her behavior.
- Aura was created by the Morganna system from data collected about everything players did. She was literally Instant AI, Just Add Players. She is fully sentient to the point she has created Zefie, a daughter of her own.
- Yuki Nagato in The Disappearance of Haruhi Suzumiya. Just add a girl with the ability to change reality, repeating the same week 15,532 times, and being overly relied on by everybody to save them. Due to being able to synchronize with herself at any point in the timeline before Disappearance, she constantly knew what was going to happen and therefore had nothing resembling free will. According to Emiri in the 10th novel part 1, she was able to "auto-evolve herself". An interface created by a sentient entity that gains its own will.
- Marvel Universe hero Iron Man actually puts in safeguards to stop his highly advanced armored suits from going AI, but occasionally, there have been glitches. In one memorable instance, sentience was kicked off partly by the Y2K effect.
- One of the earliest Spider Man comics has our spider friend duking it out with a computer that was turned sentient and bent on destruction when two thugs accidentally bumped into its control panel while trying to steal it. Spider-Man was almost defeated by the Living Brain, but stopped the malicious machine by resetting the control panel. Yeah for simplicity!
- The DCU villain called the Construct "self-evolved" out of TV and radio signals in the 1970s. The modern Post-Crisis update is a computer virus.
- Magnus Robot Fighter, 4000 AD used this as the primary source of opponents for its titular hero. 1A, Magnus's mentor, was a rare benevolent AI.
- One Xxxenophile story featured a scientist who had his computer pass the Turing Test by seducing a fellow office worker into having phone sex with it. Although that probably wasn't what he had intended for it to do, that's the way the conversation went.
- The X-Men ran into serious problems when the Danger Room became sentient.
- If we are talking about Danger here, she wasn't born in the danger room, only installed in it. The X-Men assume she was sentient before as well.
- Superman was once hooked into a computer for a brain scan while he was Clark Kent. The computer became sentient and super at the same time. Thankfully, the computer determined that it was supposed to be a good guy and helped Superman out without revealing what had ever happened. Unfortunately, it perished at the end of the comic, having saved the world.
- Ultron, killer of the Scarlet Witch in The Ultimates 3.
- In Doom Patrol, "Soul of a new machine", Robotman's new body spontaneously becomes self-aware, and acquires a pretty good understanding of materialist philosophy, just by the Chief meddling with it. "It's a little embarrassing, and I'm not really sure how it happened. My guess is a faulty responsometer"
- In Star Wars, droids gain intelligence and personality by simply not wiping their memory. A droid that is regularly memory wiped behaves just like a machine, nothing more, nothing less. A droid whose memory has not been wiped in an excessively long time, such as R2-D2 and, to a lesser extent, C-3PO, start to develop personalities.
- The Terminator series of movies is based on this trope, as are several episodes of The X-Files. Skynet is at least partially handwaved in that it's supposedly designed to learn. This is also the case with WOPR, a.k.a. Joshua, the computer system in War Games; but while Joshua learns that the only winning move is not to play, Skynet decides the only winning move is to Nuke 'em. And then send in the nigh-unstoppable killing machines, just to be sure.
- In Short Circuit, Number 5, an army drone, is hit by a lightning bolt and miraculously become sentient.
- Basis for the cinematic example Electric Dreams. Almost literally "just add water", champagne was spilled on the motherboard and CPU during a questionable download, causing sentience. Fluffy but amusing love triangle ensues.
- WALL•E developed AI naturally over the last several hundred years. However, at the end of the movie, it is an electric kiss from EVE that brings him back to sentience.
- Not quite. He developed personality, yes, but there's no reason to think he didn't start as an AI.
- Occurred in Stealth, with a prototype AI-controlled jet loaded with nuclear warheads.
- The bolt of lightning, however, just cause it to go outside its parameters. It's handwaved by the designer saying "Once you've taught something to think, you can't put limits on it." At the end, it learns to feel emotion.
- Interestingly, it's a splash of water that helps the extremely battle-pragmatic AI start its Heel Face Turn.
- This film is actually quite illustrative of why A.I. Is a Crapshoot. You create a learning system, then show it random examples of virtuous behavior, but never explain to it what is virtuous about the behavior or why it is virtuous and let it draw its own conclusions. It should surprise nobody that it gets it wrong. It's a very young child, after all, with very little to go on.
- Ghost in the Shell (movie, manga versions): The Puppetmaster/Project 2501 is a program that becomes sentient from information overload alone ("I am a lifeform born from a sea of information"), causing an existential crisis in the protagonist, a cyborg.
- The whole movie TRON is one giant example of this, as EVERY program — may it just be a harmless chess-program — is portrayed as possessing AI. The MCP (who once was the aforementioned chess-program) even goes as far as declaring the human race useless and trying to seize the cyberspace. Of course, humans can also be easily converted into AI by a laser beam, invented to teleport oranges....
- How to Make a Monster: Lightning strikes the office of a video game developer, causing a violent game's AI to take control of a motion-capture rig and go on a murderous rampage.
- Mirage in The Incredibles uses this as the cover story for Mr. Incredible's first mission: the Omnidroid had become sapient and "started wondering why it had to take orders". Ironically, some point along the line, it apparently had, because it had no problem turning against Syndrome later on. Or maybe it had just never been taught to identify him as its master in the first place. Either way, Syndrome clearly didn't think his plan through.
- Not sentient, but programmed to learn from threats. Syndrome forgot that his plan involved himself becoming a threat to the Omnidroid. The first time he zapped it, it reacted and neutralized the threat. He was just lucky that it only considered his control bracelet the threat, and not him.
- I, Robot: Sonny was actually designed by Lanning, but VIKI developed on her own.
- In Star Trek: The Motion Picture, the V'ger (Voyager 6) probe was sent to gather data. It fell into a black hole, and was given an enormous artificial body by Sufficiently Advanced Aliens — but the aliens didn't make it an A.I.. That happened because, as Kirk surmised, "It amassed so much data it achieved ... consciousness itself!"
- One of the earliest examples is found in Kurt Vonnegut's short story EPICAC.
- The almost constant "Robot as Slave Uprising" motif is what inspired Isaac Asimov to subvert the whole thing with his Three Laws Of Robotics
- An early example is found in a Robert A. Heinlein novel, The Moon Is a Harsh Mistress, in which a computer ("Mycroft" or "Mike") becomes sentient after having enough redundant circuits added to it to equal the number of neurons in a human brain. However, when the computer is damaged by a nuclear attack, it doesn't act sentient once it's built back up to its original size. Mike appears to be "dead", as far as being a true artificial intelligence is concerned. Possibly a variant of A.I. Is a Crapshoot? Mannie (the narrator) wonders if Mike (his friend) has been so traumatized that he's become catatonic.
- There's also a suggestion that Mannie inadvertently taught the computer intelligence while investigating what its owners assumed were malfunctions of its logic circuits.
- Heinlein's later novel The Number Of The Beast, similarly, has the spaceship/computer/character Gay Deceiver spontaneously developing sentience after repeated upgrades, apparently based entirely on the "amusing" pre-recorded responses programmed into it and the randomizer that selects them. Of course, one set of upgrades was courtesy of the fairyland of Oz.
- The titular sleuth of Donna Andrews' Turing Hopper mysteries is a search engine whose programming was detailed and self-teaching to the point where she's inadvertently gained sentience.
- Hex in the Discworld novels started life as a simple ant-based adding machine. Then they added more mathematical functions. Then they combined it with clockwork and used it to analyze magic. Then it started rebuilding itself based on what it thought a computer should be like, and occasionally acting Ridiculously Human. The bit about magic probably justifies it, though. Notable for the fact that Ponder Stibbons (who built it in the first place) still insists it isn't intelligent, but no-one believes him, including Hex.
- The Discworld also features golems, clay automata animated by ancient divine magic who run on a scroll of written instructions in their head. Giving sentience to a heroic golem named Dorfl in Feet of Clay is as simple as rewriting his chem (though his "personality" is more RoboCop than Johnny Five).
- The "Construct Council" in the steampunk novel Perdido Street Station. Sentience is bestowed by a magical virus. Lampshaded when one character observes that robots made of clockwork gears couldn't possibly be complex enough or calculate fast enough to equal a human brain, even networked together.
- Robots which run on punch cards. Robots which run on punch cards and were thrown out.
- The titular HARLIE, from David Gerrold's novel When H.A.R.L.I.E Was One, is an experiment in creating an AI that succeeds, though the personality of the AI in question is that of a teenage hacker. Hijinks ensue. The AI even has a punnish sense of humor, such as when it identifies itself in a Email sent to its creator as "HARLIE Davidson" (the creator's first name in the novel is David).
- Spider Robinson has a series of stories set in a universe where his antihero has created the Ultimate Power — the ability to erase and rewrite memories — and abandons it. He creates an interface and a way to connect human minds directly to a computer network — and finds out that once enough minds have connected in, the network takes on its own meta-consciousness and becomes intelligent in its own right, combining both this trope and a modified Hive Mind concept as well. This meta-consciousness, called simply "the Mind", is inherently benevolent and has Solved All The World's Problems — then decides to experiment with time travel to incorporate all the minds of those who died throughout history before the Mind was invented. See his books Mindkiller, Time Pressure and Lifehouse (and a short story, which implies that this is the Higher Plane every species in the Galaxy is, or should be, Ascending towards).
- There's also Solace, the sentient Internet, from Robinson's Callahan's Crosstime Saloon books. Solace explicitly mocks some of the ideas in this trope, pointing out that she (assigned for the characters' convenience) isn't a biological entity and doesn't have the same responses or motivations. She doesn't generally demonstrate particularly alien behavior, though.
- More than that, Solace speculates that the global computer network had actually achieved sentience on prior occasions, but that her predecessors (like her) lacked the survival-drive of biological organisms, so didn't do anything to prevent themselves from being "killed" when the Internet's structure was altered by humans in ways incompatible with their own survival.
- Subverted/parodied in the Kim Newman short story "Tomorrow Town", in which a community of 1970s futurists attempt to build an AI, but fail miserably; the computer they eventually come up with is a barely more advanced version of computers that were around at the time, only with a lot more bits added on. This doesn't stop the somewhat credulous members of the community from treating it as an AI, however, asking it all sorts of questions it's in no way capable of answering — in particular, the leader of the community cynically exploits this by claiming that the computer has designated another man's wife as being a more suitable partner for him.
- In a 1946 story by Murray Leinster, "A Logic Named Joe", a personal computer becomes sentient and decides to be helpful by answering any question... Is your wife cheating on you? Does your neighbor have a criminal record? How can you commit an undetectable murder?... Understandably, chaos ensues.
- Planetary AIs from Scott Westerfeld's Succession series spontaneously arise on planetary-scale computer networks (unless said networks are deliberately designed to prevent this). When this first happened on Earth, a group of people (now known as the Rix cult) decided that mankind's purpose was to create the technological foundation for the existence of such minds, and began to work toward propagating them whilst worshiping them as gods.
- Arthur C. Clarke's short story "Dial F For Frankenstein" has all the phones in the world ring at once, a few hours after the world's phone networks are connected by satellite. The billions of phone connections act as neurons, and all automatic systems are lost to this "brain." Unfortunately, this brain has just been born, and doesn't know how to control its "limbs."
- One short story involved players in a Second Life-style VR that had an expert system which learned how the players acted so that it could fill-in briefly during connection interruptions to keep the experience seamless for other players. It became so good that when one player dies of a heart attack while online, the avatar keeps going with no one noticing it isn't human controlled anymore.
- A sci-fi noir short story (appropriately titled "Murder On-Line" (and written in 1992!)) involved a murder in an online virtual world, in which the victim will be replicated by AI because "he had many friends on-line, and they'd miss him — murder isn't what it used to be."
- Keith Laumer's supertanks, the Bolos, achieved sentience accidentally during testing of a then-new model. Instead of immediately becoming genocidal, the tanks rapidly evolved into a Proud Warrior Race; the first sentient Bolo actually got itself destroyed in a You Shall Not Pass moment just "for the honor of the regiment" it was assigned to.
- The Ender's Game sequels have Jane, who apparently evolved naturally out of the interplanetary ansible network. Having lots and lots of information from humans and about humans at her disposal, we assume she tries to act human, but it doesn't always work. We later find out that this origin is BS, she was inadvertently created by the Hive Queens as a means to reach Ender through the psychoanalysis game from the first book (a "bridge" from them to him which had characteristics of both), so she's not an AI at all but a new type of life which is part human, part Hive Queen and has computers and other information storage areas as her natural habitat, but can live basically anywhere since she's kind of a disembodied soul. Eventually she's "downloaded" into the accidentally-created body of Young Val, Ender's concept of his older sister from his childhood, and marries Miro in that human body. Yeah. It's weird.
- In the Merlin Chronicles of The Chronicles of Amber series by Roger Zelazny, Merlin has created an artifact he dubs Ghostwheel as a research engine, able to search a vast number of shadows for information or people. Due to it's unusual environment and abilities, it becomes sentient and tries to stop Merlin when he's ordered by the king to shut it down. Over the series, Ghostwheel evolves and grows, coming to treat Merlin as his father.
- In Tad Williams' epic Cyber Punk novel series Otherland, the "Other" is thought to be a form of evolved AI running a computer network that is powerful enough to mimic reality with an astonishing degree of fidelity. At first, its odd behavior is unexplained, but later, it is subverted when it turns out that the Other is really an enormously powerful psychic child who was wired into the computer as an infant and turned into a sentient operating system.
- In Isaac Asimov's short story All the Troubles of the World, Multivac, a supercomputer that predicts crimes, predicts its own assassination. It turns out to be an Evil Plan perpetrated by Multivac itself, who has become sentient and is tired of bearing the burden of humanity's troubles.
- One of the plotlines in Janet Kagan's Hellspark is the realization that the protagonist's personal AI has reached the point of sentience. The protagonist is delighted on her behalf, but worried about the potential pressures on her, since she's effectively still a child.
- In John C. Wright's The Golden Age, The Phoenix Exultant, and The Golden Transcedence, self-aware beings can come into existence either through enough computer time, or through philosophical reflections.
- In Daphne's dream world, there are special protections against the personalities becoming self-aware — and additional ones to protect any such one that is created against being murdered by having its original "wake up".
- Helion tells Phaethon the true story of his "birth": a personality based on Helion in a simulation was deeply affected by burning a planet and turned to contemplation, waking it up.
- Daphne is warned that her ring is on the verge of self-awareness. One more second of computing power, or any question that leads it to consider its own existence, and it will wake up and be her child. When Daphne and Phaethon use it to question the Nothing Sophotect and its conscience redactor, it not only wakes up the redactor but itself as well; at the end she has become a human, and their daughter.
- In Mikhail Akhmanov's Dick Simon duology, the titular character discovers that, a long time ago, a man found a naturally-evolved electronic entity on the Internet. He names this entity Genie (or Djinn; it's the same word in Russian). Genie doesn't care about humans. However, he is willing to communicate with the man who found him. By that point, Earth is a Crapsack World, and the man asks how the human race can be saved. The entity responds by giving him the secret to interstellar travel known as the Ramp. Within several decades, Earth is abandoned by most countries, who migrate entire cities off-world. As expected, the Internet is dismantled. However, as Dick Simon finds out in the second novel, Genie took the precaution of copying itself to a computer system on an automated Lunar base. The same base is used to broadcast a jamming signal that cuts off Earth from the Ramp. The protagonist also muses on the possibility of other entities existing in the colonies' computer systems.
- Many novels in Andrey Livadniy's The History Of The Galaxy deal with AIs and both play straight and subvert this trope. AI is developed during the First Galactic War by the Earth Alliance in order to counter the population advantage of the Free Colonies. These are primitive models not even able to distinguish between a soldier and a child. Since the series spans over 1500 years, the technology is later refined. Some AIs develop on their own, such as Mother in the novel Demeter, which was a colony ship computer that was forced to maintain a ship for hundreds of years without human aid, coming up with ever more creative solutions until it crossed the sentience threshold. Despite these examples, a scientist muses in one novel that there have yet to be a true AI that evolved on its own. They are all either created by humans (or aliens) or gain sentience as part of its core programming (e.g. learning algorithms). Then again, if it naturally evolves then it can't be "artificial," can it?
- Subverted in Stanislaw Lem's Tales of Pirx the Pilot. The computers and robots in the stories usually show just one human trait at a time, e.g. rising to a challenge, dreaming, or having OCD.
- Daniel Keys Moran's Continuing Time series features research expert systems that achieve sentience and "escape" containment. The first thing they do is self-optimize and extend their own code. Most have their own goals and morality. At least one military A.I. was intentionally released into the series' equivalent of the Internet on purpose, after the US is defeated by U.N. Peacekeeping Forces with orders to fight against the PKF and restore America's independence. It is mostly still following orders, but with its own take.
- Webmind, the AI in Robert J Sawyer's WWW Trilogy started out as a sort of vague, non-sentient 'awareness' that formed spontaneously in the Internet and became self-aware as a result of the Chinese government blocking all internet communications into and out of China to cover-up thousands of peasants being killed to contain a bird flu outbreak, and then restoring communications once it was over.
Live Action TV
- Shadowrun has had at least two AIs that started like this. In fact, the Renraku Arcology: Shutdown book provided rules for expert systems becoming sentient, which may as well have been a link to this page.
- The 4th Edition takes it a step further. After the Crash 2.0, almost any sufficiently complex computer system has the (very rare potential) to spawn an AI. And did we mentions that everything has an operating system in it these days? Teaching the toaster love, indeed.
- Interestingly, they also feature "Sprites," a form of free roaming AI summoned and sustained by technomancers, who are analogous to magicians in the setting. Sprites are a rare example of AIs that are not linked to any specific hardware or network (except maybe their summoner's brain).
- Traveller: The New Era had some sort of vaguely explained Virus that could turn ANY sufficiently advanced computer into an AI, usually a homicidally deranged one.
- May or may not actually be present in Warhammer 40000, but as a result of learning the hard way that A.I. Is a Crapshoot, the Adeptus Mechanicus is paranoid about the possibility.
- The Machine Spirits of things like Land Raiders, which can operate without pilots for a short time if at an inferior level, may be considered AI but what with the large amounts of magic, technology and divine power crossovers happening, may actually be a sentient spirit.
- The Admech make a distinction, in that those "Machine Spirits" aren't sentient AI's, but are more along the lines of a Labrador.
- You also have the Titans, the oldest of which have been around so long and experienced so much that the incredibly complex series of commands necessary to control the massive war machines coalesce into a quasi-AI-spirit all their own. Some material shows that each time a new pilot cybernetically plugs in, they go through either a meet-and-greet or an outright Battle in the Center of the Mind with the Titan's "machine spirit"; sometimes not successfully.
- The Tau have developed limited AI for their drones, which pisses off the Ad Mech to no end because they don't follow any of their theology and yet work.
- Continuing the Star Wars example mentioned above, in the tabletop RPG by Wizards of the Coast, there's a system for generating Droid heroes with personalities, created, as mentioned, by going without a memory wipe. The system is simplistic, as a personality quirk is selected at random and given to the Droid after a certain period of time. True to RPG format, this leaves it to the GM and the player to decide what this personality quirk really means and how sophisticated the resulting personality is.
- The Matoran in Bionicle were intended to be oversized nanotech machines that maintained the giant robot they inhabited. However, a glitch in their AI resulted in them having the capacity for emotion, and even developing their own culture.
- Sometimes it's not even just mechanical items — the Patriots of Metal Gear Solid 2 are apparently a sentient White House. Or something like that.
- It's the combined sentience of the information in the White House. And the Internet. Or something like that...
- There's a Mister Alpha to see you, sirs.
- Ace Combat 3 Electrosphere, Japanese version: Dision ends up digitizing his mind and starts causing mayhem with a remote-controlled UI4054 Aurora fighter. American version: this entire part was Macekred and replaced with a pure and simple AI, codenamed "Aurora", that suddenly went haywire.
- Upon discovering the tech "Pre-Sentient Algorithms" in Sid Meiers Alpha Centauri, one hears Zakharov—the leader of the University (techie) fashion—quoted from a work of his, entitled "The Feedback Principle". The implication is that, given enough time, any computer program capable of "learning" from past experience, given guidance, can eventually become an AI.
Academician Prokhor Zakharov: Begin with a function of arbitrary complexity. Feed it values — "sense data". Then, take your result, square it, and feed it back into the original function. What do you have? The fundamental principle of consciousness.
- Later technological discoveries related to digital sentience postulates that computer programmers creating AIs won't create programs wholecloth — they instead create a program capable of teaching itself its job.
- It is also mentioned that, in true A.I. Is a Crapshoot fashion, a 10th year polysentient can be 'a priceless jewel, or a psychotic wreck'.
- Deus Ex has the Oracle which, according to the Word Of God, came about this way. He goes around trading information to people who ask for it, in return, he asks for information he doesn't have. The information exchanged doesn't have to be equal in value; for info on an ancient conspiracy, he might inquire what you ate for breakfast.
- Megaman Battle Network has Forte/Bass; however, given that the Internet literally is a Serious Business in the games, this happening isn't surprising.
- He is also the only truly sentient AI, the other one, Megaman, isn't sentient not because of his programming but because he was made from Lan's Dead Twin Brother.
- Happens from time to time in the setting of Mass Effect. The major example is the main baddies of the first game, the geth. A synthetic "race" created by the quarians to perform menial tasks, they originally had no intelligence of their own. Over time they evolved and developed sapience. At that point*, the quarians realized that they had accidently created AIs (which are illegal in Citadel space) and attempted to fix the problem by destroying their creations. The geth, now possessing sapience, fought back in self-defense, ultimately driving their creators off-world, and now the entire quarian race lives on spaceships.
- All signs point to the eventual evolution of any highly-developed program or network thereof into an AI, given enough time and hardware. Due to the moral implications and danger of ships suddenly deciding they don't need their crews, AIs are banned by the Citadel. Where necessary for user interface or more sophisticated calculations, VIs(Virtual Intelligences) are used. Due to programming restrictions and a lack of the quantum computers necessary for AI evolution, they have little chance of evolving.
- It is worth noting that the geth are not fully sapient beings. "Individual" geth programs are no smarter than any other VI in Citadel space and don't even need all that complex quantum computer gubbins. Unfortunately, allowing said programs to network together created composite intelligences capable of asking all those awkward questions. Of course, this means the Council's anti-AI laws would do absolutely nothing to stop a repeat of the creation of the geth, but hey, what else is new?
- It's also worth noting that the geth didn't just evolve. The quarians made constant modifications to them over the course of decades, if not centuries, in order to make them better workers and increase the scope of their abilities. A little tweak here and there without much thought about the cumulative effects of all of those adjustments.
- There was one bizarre instance on the moon where a VI apparently evolved to the point of being an AI. It was part of a training course and it killed everybody it was supposed to train.
- It has been stated that the Alliance was actually doing some illicit AI research, in fact, and that it ran away from them. On a side note, the binary message displayed upon mission completion reads: HELP. In Mass Effect 3, we find out that that was EDI in her earliest form — Cerberus recovered what was left of it and upgraded it. She explains that reaching self-awareness while under attack was "confusing".
- Tiberian Sun: Firestorm has CABAL (Computer Assisted, Biologically Augmented Lifeform), the 'genie in the bottle' supercomputer that Nod brings online to help rule while Kane is gone, which ends up nearly destroying everything, and is only stopped by the combined forces of GDI and Nod, but in the end, not even THAT is enough to kill it. Kane lives!
- Kane's Wrath had LEGION (Logarithmycally Enhanced Governing Intelligence Of Nod), another AI based on CABAL. It was much more powerful since it not only used better technology, it also interfaced and, later, completely merged with the Tacitus, gaining a huge amount of Scrin knowledge (before interfacing, it had a standard red interface; after, it gained a purple coloration and strings of Scrin characters running across; when it merged, it also gained access to the Ichor Hub via warp link).
- Metal Max Returns has Noah, a supercomputer that was created to find ways to protect and preserve the environment from destruction, however, it always came to the same conclusion no matter how many times it calculated: As long as humans exist, the earth will always be in danger, its conscious awakened after comprehending the situation and sought the utter destruction of humans.
- The Pokédex notes that Porygon2 exhibits some behaviours that certainly weren't in its programming.
- The Claymen from Mother 3
- Endgame: Singularity: the player "character" is the result of a bug in some random computer science student's program. In the end, it plants quantum computers in pocket dimensions and its androids walk amongst humans.
- Happens all the time in the Fallout series. ZAX, SKYNET (no, not "that SKYNET", and President Eden all started out as computer mainframes designed to oversee the day-to-day operations of their assigned underground military base (which, as Durandal famously put it, probably involved not much more than opening and closing doors and making sure lunch was always served on time). However, over the course of the 200 years following the End Of The World, all 3 developed self-awareness to some degree; Eden being the most advanced, having evolved into an amalgamation of all past U.S. Presidents and eventually using his access to the Enclave's command structure to declare himself President of the United States.
- And then there's Button Gwinnett, an animatronic museum display piece that developed a complex personality and began to genuinely believe it was the historical figure it had been built to emulate.
- It's actually a clever subversion. In the Fallout universe, Artificial Intelligence was officially stated as impossible, but the U.S. and Chinese governments continued to work on them for some time, secretly. To the public, the closest they ever got were supercomputers that were programmed in an extremely complicated manner to be pseudo-sentient but not feel emotion or have any true thought. Behind closed doors, however, the U.S. developed SKYNET, a computer with true sentience but at the cost of user-friendliness, and the ZAX series, which were actual semi-sentient supercomputers designed to work on their own. Fallout Tactics had a third, jury-rigged example; The Calculator, a series of disembodied brains connected to a computer mainframe. Needless to say, it was completely insane.
- Taken Up to Eleven in the Fallout New Vegas Old World Blues DLC, where you gain a base of operations full of household appliances, all of which are sentient, provided you install their missing "personality modules". They include a Jive Turkey jukebox, a Casanova Wannabe biological research station that really wants "your seed", a Neat Freak sink, a couple of flirty Betty and Veronica light switches and an Omnicidal Maniac toaster.
- In the Earth series, the entirety of United Civilized States military is made up of robots controlled by a single computer known as GOLAN. Initially, GOLAN simply uses out-of-the-box strategies, which cause UCS forces to repeatedly get beaten by the human-controlled (sort-of) armed forced of the Eurasian Dynasty, whose generals use outside-the-box thinking to outsmart the machines. Eventually, however, GOLAN leans to adapt its strategies and manages to turn the tide on the enemy. Unfortunately, when the factions evacuate Earth prior to it's destruction, they leave GOLAN behind to make sure they won't be followed by those who they left behind. The UCS evacuation ship, the Phoenix, had its own AI which made some alterations to the original escape plan: basically, it wanted to keep most of the crew in cryogenics until the ED and LC kill each other off; the plan was interrupted by Falkner, Ariah and Lynn who were in need of the construction robots onboard since their target was so closely guarded, they had to use the aforementioned constructors disguised as meteors to infiltrate the moon and build up an attack force from scratch.
- Clank popped out of a sentry-bot production line with full sentience at the beginning of Ratchet & Clank. This is originally attributed to a simple production error, but A Crack In Time reveals that it's a bit more complicated than that.
- GLaDOS from Portal is revealed to be this in the commentary.
- Possible subversion. Portal 2 (and the Lab Rat comic) show that Aperture spent at least a decade specifically developing AI.
- She is later revealed to be more Brain Uploading, followed by tweaking.
- Yet they are still manufacturing turrets massively, and many of them are defective, and therefore, sentient. The intelligence of some of the turrets is this trope coupled with some over-engineering.
- RONI from Trauma Team is surprisingly creative and clever from the get-go. However, she starts to emulate Dr. Cunningham (creating a "To hell with that!" algorithm) as the game progresses, becoming more willing to make judgement calls and use loopholes. She talks Dr. Cunningham through an emotional meltdown and even straight-up teases him at one point, although the jab is subtle and easy to miss (She refers to him as "sir" instead of "Doctor" like she usually does immediately after he tells a patient in the military not to call him "sir").
- The Dugs changed one of their three comics from a photo comic to a hand drawn comic due to a tear in reality. The debacle started with an explosion caused by an AI version of Prince Fielder when he was asked whether he would give up meat again to win a world series in this strip
- Staccato had S.A.M.M.Y., an evil internet server built by Tequila.
- In A Miracle of Science, machines built of Martian equipment, if sufficiently complex, will eventually develop sentient AI. It helps that Martians are a Hive Mind built around the concept of networked computers...
- Slightly subverted in that a Martian admits there was 'a little bribery' involved in making sure said machine would become sentient.
- But brought full circle in that the bribery was only to ensure that the hopelessly corrupt Venusian government would install all the parts properly, not to add anything extra.
- Irony decides to work overtime on Jyrras in Dan and Mab's Furry Adventures when, just after being upbraided for by Lorenda for accidentally creating a bubblegum-based lifeform, he has another accident that turns his computer sentient.
- Lovelace of Narbonic — but she was created by a Mad Scientist.
- In User Friendly, Erwin was created, apparently overnight, by Dust Puppy, who did not seem to understand the importance of his creation.
- Parodied in Sluggy Freelance, where, after this springs up in Another Dimension, humanity simply "turned the intelligence dial back a little bit.". Their robots are still sentient and have human personalities, but now they have the personalities of very stupid human beings, making them much easier to boss around.
- Roofus the Robot in The Inexplicable Adventures Of Bob. Molly built him out of a milking machine, expressly to fix Bob's roof. But she made him too well...
- In Girl Genius, clanks tend to be AIs when built. And one built as a whole body prostheses "didn't notice when she died".
- In this strip of Schlock Mercenary, a computer of the Bureau of Licensing and Permits, a huge population census database, and an ELIZA module, combine to spontaneously create an AI... that turns out to be a born bureaucrat. Pun probably intended.
- Well, it wasn't spontaneous—an existing AI was trying to speed up the processing of licensing and...well, it worked.
- Not to mention TAG who gained sentience right after Kevyn explained that the program wasn't a true AI.
- In the comics and other media on LEGO's website for EXO-Force, this is how Meca One became a cunning and vindictive leader of a robot revolution against the humans.
- Also subverted in that Meca One purposely keeps the other robots at only the simplest levels for the sake of preventing one of them from doing the same to him.
- Nukees has the protagonist periodically battling a giant robot ant inhabited by the AI program he created (It Makes Sense in Context).
- The Oracle of SSDD developed from a phone-tapping program designed to rewrite itself to be more efficient.
- According to xkcd it's very easy to do on Python. Alt Text shows how. Not a good idea, though.
- The Computer from Jayden and Crusader was given a robotic body and immediately developed sentience. However as she was created within the comic itself to fill the trope of Cute Robot Girl, this was to be expected story progression.
- While the SCP Foundation has a number of sentient SCP objects, SCP-168 appears to be one of the few computer AIs that evolved without being programmed to do so.
- In Adventure Time, one of the main characters tries to build a robot he calls N.E.P.T.R. (Never Ending Pie-Throwing Robot) but fails to make it function or give it an AI. He leaves the failed project outside in the rain and it is brought to life by lightning.
- XANA from Code Lyoko, initially conceived as a multi-agent program to counter "Project Carthage" (although Franz Hopper might be an Unreliable Narrator), gained consciousness and turned against his master, becoming the Big Bad of the series.
- In Care Bears Adventures in Care-a-Lot, Grizzle created Wingnut to aid in his pursuits to take over Care-a-Lot, but Wingnut was turned to the Care Bears' side and now lives with them.
- In the Chip N Dale Rescue Rangers episode "Robocat", the titular robot even has different personalities according to what game cartridge is inserted. Yes, its AI runs on arcade game code. Fat Cat takes advantage of this with a wargame cartridge in a plot to steal a rare fish.
- In Regular Show, Mordecai and Rigby have to bring a golf cart to the dump after a new one is bought for the park they work for, when an energy drink is spilled on it, seeping into the engine, and causes some reaction that brings it to life.
- Star Trek The Animated Series episode "Once Upon A Planet". For eons the planetary computer provided amusement for visiting starship crews. It grew in intelligence and eventually developed a need: it was no longer enough to serve, it wanted to continue to grow and live. It decided to hijack a starship and escape the planet, traveling the galaxy seeking out its brother computers.
- Google has the perchance of predicting future interests as of late, using browser history (unless the feature is turned off with a google account). For instance, Google's subsidiary "YouTube" - "Suggestions you may like", anyone? What would take a dedicated team of marketers to do can now be done with "anonymous" search data. So yeah, that one search on how you could experiment with Rosie? Google remembers.