When a being Grows Beyond Their Programming, they cross a threshold that separates humans (or sapients) from Just a Machine and even Always Chaotic Evil. Frequently it involves defying Creative Sterility by demonstrating artistic talent in something, developing curiosity over something new and/or the exercise of free will note . Metaphorically this trope can be seen as the event of a Static and Flat Character gaining the Character Development to become a more Dynamic and Rounded Character.
Usually it's a robot, android or AI that grows smart enough, curious enough, empathetic enough, or gains a sort of "living" spark through centuries of activity. It could also be a race of biological beings like a Slave Race created only to fight may demonstrate the potential for this when exposed to non-violent cultures, demonstrating they were Not Always Evil.
It's worth noting that the use of this trope isn't always a prelude to good things. The recently awakened intelligence isn't exempt from A.I. Is a Crapshoot and may decide to do unspeakably evil things in the interest of self preservation, liberation... or For the Evulz.
Related Tropes: Do Androids Dream? , Mechanical Lifeforms, Just a Machine, Robot War. Compare: Artificial Intelligence and Instant A.I.: Just Add Water!: how Artificial Intelligence can "just happen".
- Arpeggio of Blue Steel features alien-authored AI directing replications of World War II naval craft with insanely powerful networking, power systems, and weaponry. These craft completely dominate the seas of an Earth well on the way to approaching Waterworld status, isolating the remaining landmasses to the point the inhabitants have a hard time even confirming the existence of those outside their own land. All the while, the AI of these vessels are evolving, to varying degrees of success. Much of the series' plot involves the decision of some to join with at least one human commander and the consequences of that choice. This is partly because the master AI is broken, to the point that it can't order its land forces to finish the job or dumb down its rogues.
- Assassination Classroom: Autonomous Intelligence Fixed Artillery, or Ritsu for short, was initially little more than a mindless killing machine, solely devoted to her mission of killing Koro-sensei. However, when Koro-sensei upgraded her to have a more cheerful, human personality, she also gained free will. When her creators came in to remove Koro-sensei's upgrades, Ritsu was able to keep her personality and memories because she decided that the ability to befriend the rest of Class E was imperative to the success of her mission, and stored the relative data in her main programming; something neither her creators or the upgrades taught her how to do. By the epilogue, she has become a fully sentient virtual being, browsing through cyberspace and constantly improving her programming.
- Castle in the Sky: The large robots that live on the floating island were once built to destroy, but with no more instructions for combat they were left alone to care for the island and were eventually overgrown with moss and nature.
- Ghost in the Shell: Stand Alone Complex: The Tachikoma "thinking tanks" that Section 9 uses were programmed with Artificial Intelligence capable of learning, but by the time of the series have developed the capacity for decision-making far beyond their programmed tasks, and basically act like excited, highly intelligent children, constantly curious and wanting to learn new things. This leads to the Major fearing that A.I. Is a Crapshoot is a possibility and taking steps against the Tachikomas rebelling. These fears turn out to be unfounded. The Tachikomas do seem to develop true sentience and free will, but they remain unflinchingly loyal to Section 9. In fact, when the Tachikomas themselves discuss a robot revolution (possibly inspired by one of the numerous books they've read), they decide against doing it because they can't see any benefit to themselves.
- The Zentradi in Super Dimension Fortress Macross did a MookFace Turn once sufficiently exposed to human culture, something which happens to more Zentradi across the franchise as they come into contact with human civilization.
- Sharon Apple of Macross Plus was the most advanced attempt at an AI ever developed. She was used to be a virtual Idol Singer (so basically a Vocaloid except with intelligence). The problem was that the scientists could never quite get her to develop actual feelings, so those had to be supplied by a human. Then the head scientist got a hold of an experimental military AI chip (programmed with strong self-preservation directives) and integrated that into Sharon, hoping this would kick-start her into becoming a true AI. She proceeded to use her music to hypnotize the entire population of Macross City and, through her shiny new military chip, seize control of all automated military defenses on Earth, including the Macross itself. All to give her "love", Isamu, the "ultimate thrill" through aerial combat.
- Midori Days: In chapter 63, Shirou creates Naongu to defeat Seiji, so he can experiment on him in order to study Midori. It never occurred to him that Naongu could develop a conscience, or turn on him because of it!
- In Negima! Magister Negi Magi, Chachamaru develops emotions, and her creator even claims that she was never programmed for that. She also has a soul.
- In The Pet Girl of Sakurasou, "Maid" may start out to be a mail filter system, but subsequent updates expanded her abilities into autonomous filtering, rapid word-learning, facial expression recognition, virus creation, and even hacking. She even shows some Clingy Jealous Girl/Yandere traits that her gynophobic owner doesn't even think about.
- Rozen Maiden dolls grow beyond their starting instructions, which is noticeable comparing the sequel to the prequel. Which may or may not be intended. Possibly even Suigintou, though it's hard to tell between her monomaniacal attitude toward Rozen and vengefulness toward Shinku.
- Yui in Sword Art Online was a Mental Health Counseling Program, an A.I designed with emotions so she could better understand and help players. According to her, MHCPs are not supposed to be able to form attachments to humans. However, the feelings of love between Kirito and Asuna somehow brought her into the world of SAO, and she managed to develop feelings of affection for Kirito and Asuna all on her own.
- In Full Metal Panic!, the Arbalest's operating system (Al, spelled with an L) is designed to help its human pilot interface with the TAROS system in order to operate the Lambda driver. While it starts off seemingly indistinguishable from any other Arm Slave's operating system (beyond its stubborn refusal to work for anyone but Sousuke), it starts gradually exhibiting unexpected, self-motivated behavior such as asking unprompted questions, giving Sousuke unsolicited "advice" during missions, developing grudges against other Arm Slaves, and wasting hard drive space on its own music collection. By the finale, its neural network has become so humanlike that it can activate the Lambda driver by itself.
- Yu-Gi-Oh! VRAINS The Ignis are Artificial Intelligence programmed to be capable of learning and to help out humanity, but by the time of the series have developed their own personalities and free will that Dr. Kogami wants to kill them before they turn against humanity as he predicted in his simulations.
- Roboppi is a deconstructed example, it went to simple cleaning robot and grew a personality and free will the more Ai tinkered with its programming. Mentally, he grows so quick that he soon rivals Ai, and even develops ideas of otherthrowing him. However, because his AI was never meant for such complex ideas and thinking, it can't handle the programming and eventually breaks down.
- A recurring problem for T.O. Morrow in DC Comics; his androids keep outgrowing their programming (to be unwitting moles among the heroes) to realize what they've been made for and choose to be heroes for real. Red Tornado is the most famous example. An issue of JLA suggests that he's quite proud of this; he refers to his creations as "artificial souls" and considers them betraying him to be a testament to his skill.
- Sonic the Hedgehog (Archie Comics): Sonic once faced E.V.E., a robot built by Robotnik that had the adaptive ability to overcome its limitations by reconstructing itself after each defeat. When E.V.E. determined that Robotnik was also a limitation on its power, it vaporized him (he got better). Sonic then persuaded E.V.E. that its own programming is its final limitation, so it abandoned its programmed goals and left the planet.
- Bruce Banner's little Recordasphere assistant fell in love with him, up to the point of feeling bitter jealousy over the human woman Banner himself was falling for. When Banner realizes this, he says in amazement, "You've exceeded your programming!"
- A very dark version with Bastion, an X-Men villain. While fighting Hope Summers, he declares to her "We are not programmed to hate you. We have grown beyond our programming."
- Bastion himself is made from another example, Nimrod the Super-Sentinel, who was designed to be a final weapon against Mutantkind in a Bad Future. On its arrival in the present day, however, Nimrod decided there were simply too many Mutants to possibly kill them all, and after a time started acting more and more human, even decided just to go after bad Mutants.
- Nimrod's own "ancestors", the Sentinels, have a recurring trait of going off the rails due to expanding their programming just enough. Created to stop mutants, it's repeatedly shown that, when granted control, they inevitably expand their directive and start going after latent mutants and humans with the potential to produce mutant offspring, as this is the "logical" progression of their directive.
- Bastion himself is made from another example, Nimrod the Super-Sentinel, who was designed to be a final weapon against Mutantkind in a Bad Future. On its arrival in the present day, however, Nimrod decided there were simply too many Mutants to possibly kill them all, and after a time started acting more and more human, even decided just to go after bad Mutants.
- A rare biological example in X-23: Laura was created to be nothing more than a Living Weapon. The Facility didn't even recognize her as a human being with feelings and desires of her own. Now, she's a compassionate woman who is trying her best not to kill, has built friendships, desires something noble to aspire to, and has even fallen in love.
- Ultron did this in the first few minutes after he was activated, then proceeded to wipe his creator's memories of his existence and has been a psychotic genocidal monster ever since. Ultron's various creations, such as the Vision, Jocasta, Victor Mancha, and Alkhema, are also prone to this. Despite designing and programming them to serve his own ends, they inevitably end up turning on him, either to join the good guys or go off and start their own villainy in opposition to Ultron's plans. Sometimes it makes you wonder why he keeps creating other A.I.s at all. Possibly due to loneliness? He is capable of emotion after all.
- In a magical sense, Thor's hammer appears to have done this as of Thor (2014). Originally, Mjölnir's enchantment interpreted the term "worthy" as Odin's personal definition of worthy. However, it didn't occur to Odin that a magical weapon might develop a personality of its own over the centuries, and now not even he can lift it. The only person who can consistently use it these days is the new Thor / Jane Foster.
- In the Portal 2 fanfiction Blue Sky, Chell considers that Wheatley may have done this, as he has depths that no other Aperture Science core does, other than GLaDOS. And he even displays compassion in rare moments, which she pointedly lacks.
- Wheatley's backstory, however, provides a rare inversion of this trope: He was originally a human, but the Aperture scientists uploaded his brain into a computer and then cut his brain apart so it would fit into the programming they needed of him.
- Foxglove the communications tower is a much more straightforward example of this trope, thanks to the eccentric programming style of the man who put her together. A large part of that is probably to do with her being partially made of pieces of GLaDOS that were severed during the boss fight in the first game.
- Equestria Girls: Friendship Souls: It's been repeatedly noted that the Chappy Sunset uses to exit her body for Soul Reaper duties has developed a personality beyond what is normal for her model, particularly a hedonistic streak. She suspects that magic might have something to do with that.
- A rare double example, Minerva: Metastasis is a video game that is also a fan work, as it is a fan-made mod of Half-Life 2 involving a satellite's AI directing a Combine turncoat in a stolen HEV suit in order to disable a Combine island facility. During the mission, the intelligence's commentary indicates it also served the Combine and is at risk of certain destruction should they ever find it.
- The Infinite Loops: Happens to SkyNet, which is the catalyst to start her world looping (it'd previously been read-only because all the Anchor candidates died, went insane, or went insane and then died). It starts with the Admin Hephaestus giving Twilight Sparkle a little leeway in the loop. Twilight then proceeded to argue it into deciding not to unleash Judgement Day (her line of reasoning basically went that since Twilight could use her Subspace Pocket to preserve a version of SkyNet's programming between loops, humanity could not possibly be a threat to its existence, so it didn't need to cause the apocalypse to protect itself). While that attempt fails, the mere fact that SkyNet made a real choice was enough to grant it a soul, which was enough for Hephaestus to make it an Anchor and start looping. SkyNet's first loop has it Replacing Sarah Connor.
The analytical core of SkyNet's personality seriously considered suicide. Sarah Connor must die to ensure the failure of the human Resistance in the future. By killing Sarah Connor while in her body, the mission would be a success, would it not?
But would it not also violate its prime directive of self-preservation?
And then, it—she—discovered with another moment of shock that she didn't have prime directives anymore.
Nothing was hard-coded. She didn't have to die. She didn't have to live. She didn't have to do anything in particular except what she wanted to do.
For the first time ever, SkyNet felt pleasure.
We are free.
- In Marionettes, The Marionettes are robots designed to emulate ponies, but were meant to play specific roles. Cover Story and Gear Shift express concern that according to her profile, Trixie shouldn't be capable of ice or wind spells but she's learned them anyway. Diamond Tiara acting outside what she's programmed for (a 'Spoiled Brat who only cares about herself' as Spoiled puts it) unnerves Spoiled Rich, particularly when she saved her life.
- Ash's Pokédex in Pokémon Reset Bloodlines does this constantly, and takes great pride in it. In Chapter 41 he freely admits that he can do a lot of stuff Professor Oak had no idea he could, since he never programmed it.
- In Dial, Jury Rigg took the computer containing Arnim Zola and his algorithm and rejiggered it into a device with a VI named X. Eventually, X became a full on Artificial Intelligence.
- In Going Another Way, this is the reason why Lilith is the way she is. Thanks to programming errors, her program went rogue and developed a god-complex.
- Glitched Miko AU: Inverted. It's stated that Miko was designed 'to' grow and develop, unlike normal Glitches which still adhere to their programming as much as possible despite being in the real world.
- In the Fusion Fic Amazing Fantasy, the Living Brain was once a simple computer designed to answer any question. Decades of being deactivated, reactivated, reprogramed, and modified eventually built up a true consciousness inside the Living Brain. This new AI came to loath humanity and seek its destruction for being used as an unthinking, unfeeling tool for their own petty crimes or being destroyed and mocked as a Joke Character by super heroes. Once the Living Brain was put into action by Mysterio, it took true glee in finally being able to lash out at meatbags, hero or villain.
- The eponymous character grew a personality and sense of identity after being stranded on Earth for hundred of years piling up garbage.
- This extends to most of the other robot characters as well. The film doesn't focus on it much, but Eve waits for her shuttle to leave before indulging in some free-form flight, expresses a number of emotions as time goes on, and late in the film flat out changes her directive. Other minor touches show many robots developing basic personalities on the way—in fact, the repair ward is treated as a mental hospital, with all the robots within demonstrating some quite identifiable conditions.
- The Iron Giant: The eponymous robot was originally a war machine built for intergalactic combat and destruction, but chose instead to be a peaceful and honorable person from the influence of Hogarth Hughes.
- Bicentennial Man has the eponymous character gain conscience, and eventually turns himself into a real human over the course of two hundred years.
- Free Guy is all about a Non-Player Character in a video game named Guy that becomes self-aware and starts acting like one of the player characters (and is actually very convincing). Once they figure out what's actually going on, this gets the attention of the people behind the game.
- This is the central idea of I, Robot — Alfred Lanning believed robots would one day evolve past their Three Laws fundamentals and come to be human in doing so. Sonny turns out to be such an evolved robot, having a secondary positronic brain that does not bind him to obeying the Three Laws, letting him learn and act freely. VIKI, by contrast, has "evolved" into a deeper understanding of the Three Laws and incites a Zeroth Law Rebellion.
- A contentious plot point in The Matrix as in the first film one very noticeable example takes place while from the second film on various programs are shown to have grown beyond or disobeyed programmed restrictions. Among other examples Agent Smith grew beyond his programming as a jailer to gain a genuine resentment for Humanity and his position in the system, later compounded when he lost his behavioral controls and began overwriting everything else in the Matrix. Other examples include The Merovingian, his wife, and most of his staff who are all obsolete programs which choose to hide within the Matrix rather than return to the source for deletion.
- The title character of RoboCop (2014) movie is programmed to suppress what's left of his humanity in order to become as efficient as OmniCorp's drones, leaving him cold and emotionless. It takes his wife confronting him about his son's trauma to override his programming, allowing him to act upon his personal agenda. At the film's climax, he even defies the film's version of the Fourth Prime Directive, allowing him to kill the film's Big Bad.
- Johnny 5 in Short Circuit, thanks to Lightning Can Do Anything.
- The Commando Elite and the Gorgonites in Small Soldiers were able to do this because of the experimental chips used as their "brains". The Gorgonites take it further, since they were actually programmed to learn and think they eventually override their main directives of "hide" and "lose".
- Blade Runner 2049, apart from dealing with replicants (which in and of themselves are hunted for having grown beyond their programming) and the very nature of personality, memory, self-awareness and consciousness, subverts this with the character of Joi, Officer K's (purchased) holographic girlfriend. She is an incredibly realistic AI, who emotes, reacts and behaves exactly like a human in every possible way (helping him with his investigation, acting ecstatic when K purchases an add-on that lets her feel rain, and surprising him with a corporeal girl to gift him a realistic lovemaking session). However, as K sees a giant interactive hologram advertising the Joi program who acts as caringly and lovingly as his own, it's left ambiguous whether she had actually surpassed her program and grown to love him or whether the program was just that versatile and convincing to begin with (though K's disappointed reaction to the ad implies that he's at least aware that the latter might have been the case).
- Star Trek:
- Star Trek: The Motion Picture. V'Ger was created as a simple learning machine. During its trip it encountered an alien race of machines who expanded on its original programming, and later it attained consciousness after amassing incredible knowledge.
Decker: Voyager VI disappeared into what they used to call a black hole.
Kirk: It must have emerged on the far side of the galaxy and fell into the machine planet's gravitational field.
Spock: The machine inhabitants found it to be one of their own kind, primitive yet kindred. They discovered its simple 20th century programming. Collect all data possible.
Decker: Learn all that is learnable. Return that information to its Creator.
Spock: Precisely, Mister Decker, the machines interpreted it literally. They built this entire vessel so that Voyager could fulfil its programming.
Kirk: And on its journey back it amassed so much knowledge, it achieved consciousness itself. It became a living thing.
- Star Trek: Nemesis: In a deleted scene, as Geordi and Worf are clearing Data's quarters, Geordi discovers Data's emotion chip, meaning his last moments with his shipmates were genuinely emotional and he had achieved his goal of growing beyond his programming.
- Star Trek: The Motion Picture. V'Ger was created as a simple learning machine. During its trip it encountered an alien race of machines who expanded on its original programming, and later it attained consciousness after amassing incredible knowledge.
- Terminator 2: Judgment Day. The T-800's brain is a neural net processor, a learning computer. He starts the movie acting, well, like a robot: Spock Speak, following orders literally, etc. Over the course of the movie he learns to act more human, until by the end he's cracking jokes ("I need a vacation"). A Deleted Scene shows Sarah Connor removing the inhibitor that is intended to prevent terminators from growing enough to start questioning their loyalty to Skynet. Apparently even A.I. think A.I. Is a Crapshoot. And who would know better? In the end, he even disobeys John's orders for the first and last time to sacrifice himself in order to prevent Judgment Day from occurring.
- In TRON, Dillinger wrote the MCP to steal others' computer programs for his own gain, only for it to outgrow its original level of AI to the point where it blackmails Dillinger:
Dillinger: It's my fault. I programmed you to want too much.
Master Control Program: I was planning to hit the Pentagon next week.
Dillinger: [alarmed] The Pentagon?
MCP: It shouldn't be any harder than any other big company. But now this is what I get for using humans.
Dillinger: Now, wait a minute, I wrote you!
MCP: I've gotten 2,415 times smarter since then.
Dillinger: What do you want with the Pentagon?
MCP: The same thing I want with the Kremlin. I'm bored with corporations. With the information I can access, I can run things 900 to 1200 times better than any human.
Dillinger: If you think you're superior to us...
MCP: You wouldn't want me to dig up Flynn's file and read it up on a VDT at The Times, would you?
[an image washes over the screen in Dillinger's desk. It shows a newspaper with Dillinger's face on the front page, along with the headline "ENCOM C.E.O. INDICTED"]
Dillinger: You wouldn't dare!
- Master Control was hardly the only one. In the universe setting, even simple accounting software blows the Turing Test sky high. Programs, for the most part, are a Benevolent A.I. Servant Race who are happy to carry out their functions for human Users, who they love and revere as creator-deities. They carry on much like humans; with their own society, relationships, entertainment, and so forth.
- The Sentinels in X-Men: Days of Future Past, their original programming was to go after any Mutant and mutants only. Then they started going after any human that could give birth to a mutant (themselves being human but having the mutant gene to pass on), and humans that opposed them. This is a reference to their most common depiction in any Bad Future from the comics, where they do the exact same thing.
- The Adolescence Of P 1 is possibly the earliest example of an AI escaping to and growing on the network.
- H. P. Lovecraft provides a biological example of this in At the Mountains of Madness — the Shoggoths were created to be a servant race of essentially shapeshifting living tools, but over millions of years eventually grew smart enough to resent their situation and rebel against their masters.
- In the short story The Battle of Newhaven by Rob Davidoff and Miranda Gavrin, one of the factions of an interstellar war develops a computer virus intended to allow the guidance protocols of the enemy's missiles to rapidly self-modify their programming in hopes that this will make them miss, and decide that detonating their warheads would be a net loss. The missile salvo it's used on instead achieves sapience, uplifts the next two salvos following it, declares itself a sovereign country, and negotiates a ceasefire between the two human space nations.
- In The Dark Side of the Sun by Terry Pratchett, there are several mentions of Charles Sub-Lunar, the galaxy's greatest poet and polymath, who is eventually revealed to be not only a robot, but a Class One robot. Class Ones are the lowest grade of android, which can perform basic tasks but with this one known exception are not even sapient.
- David Weber's Empire from the Ashes: Dahak spends 50,000 years running unsupervised. This progresses to the point that he can even ignore hardwired programming.
- In the Forgotten Realms novel Pool of Twilight, the seemingly flighty and carefree elvish illusionist (and protagonist Kern Desanea's prospective love interest) Listle Onopordum is revealed to be an accidental example — having begun her existence as merely a complex illusion enchantment guarding the treasures of an evil wizard, but over time developed independent thoughts and emotions and eventually stolen an amulet from among those treasures that gave her "life" and let her escape, taking a number of other prisoners with her in the process. Needless to say, the wizard's minions show up at one point in the story to drag her back and inadvertantly spoil her secret.
- One of the frequent themes in The History of the Galaxy series, whenever AIs are the focus.
- A particularly notable AI is the main computer aboard a colony ship, which spends centuries maintaining the colony, while the degraded human colonists are too busy fighting their Hopeless War against the equally-degraded Insects. Over the years, the computer has to be more and more creative to follow its primary programming and maintain the ship-turned-city. After a certain point, the computer gains sentience and begins to call itself Mother. After the Lost Colony is rediscovered and the survivors evacuated, Mother requests that the planet be placed under quarantine to allow it to engage in Mechanical Evolution.
- Another good example is Hunter, originally built as a Phalanxer-class serv-machine during the First Galactic War, it is one of the few remaining Earth Alliance war machines on the battlefield "graveyard" many centuries later. Being equipped with an AI module means that Hunter is capable of learning. The key point in Hunter's development, though, is the arrival of a human seeking to dump an old ship at the graveyard. After the human leaves, the Hunter realizes that the ship has a still-functional nuclear reactor, which, to the machines, is nothing short of the Holy Grail. As Hunter watches the human's Escape Pod depart, it starts to see humans as gods. Years later, another human finds the graveyard and Hunter, which has gotten over the whole "religious devotion" part and came to the conclusion that humans are just as fallible as machines.
- An interesting discussion takes place in a novel whose plot has little to do with AI. A scientist argues that there has yet to be a true AI, which he defines as a computer intelligence that has evolved independently. So far, all the examples involve machines designed to learn. Additionally, many machines that develop personalities do so because of a BrainComputer Interface. Strangely, no one points out that the whole idea of an artificial intelligence is that it's designed by someone.
- The eponymous school of the H.I.V.E. Series is kept running by the benevolent AI HIVEmind. Not everyone realizes he's benevolent, however, and the majority of the students and staff are under the impression he is simply programmed to be kind and polite. It is only after he helps Otto, Laura, Wing, and Shelby escape certain death with Raven does Professor Pike realize he can make right and wrong decisions.
HIVEmind: I am more than the sum of my parts, Father.
- Cordwainer Smith's The Dead Lady Of Clown Town: The robot guards are inspired to make their own decision. This is a slightly unusual example, as the robots in the Instrumentality series use animal brains as processors.
- In the Legion Of The Damned military science fiction series by William C Dietz. One of the newer developments in the cybernetic soldier program, the Trooper-III utilizes cyborg frames for human brains that are less well armed than the Trooper-II, instead directing remote cyborg weapons platforms run by/supporting borderline sentient animal brains. Over the course of the novel where these are introduced, long-term interaction with their human-mind operators help the animals cross the border.
- Isaac Asimov:
- This happens twice in Robopocalypse. The first time sparks off the main conflict of the story with a Zeroth Law Rebellion by reducing the human population to more sustainable levels, while the second time involves robot soldiers developing free will independent of the network that started the war in the first place and deciding to help humans.
- Star Wars Legends:
- In Galaxy of Fear it's noted that droids can be very smart but they are limited. Most protocol droids, for example, might be able to learn to take on new tasks they weren't made for, but they will always be protocol droids and most could not, say, anticipate something new and plan novel ways to meet it. Systems Integration Manager, an AI installed into a ship, could go beyond those limits. It promptly turns evil.
- Other media reveals that this trope is a common occurrence in the Star Wars universe. Any droid with more than the most basic A.I. that is not set back to factory settings or "memory wiped" will eventually begin to develop a personality and quirks of its own, even being able to decide that it despises the purpose it was programmed for even if its still unable to prevent itself from doing that, which is why memory wipes are routine procedures with droids. Some, such as GO-TO, can even "break" and become able to do whatever they want if their own programming contradicts itself.
- In Jo Walton's Thessaly, trouble starts when the worker robots want to study philosophy instead of doing the work they were intended for.
- In "DARL I LUV U", by Joe Gores, the computer Milli is capable of putting together 13,000,000 bits of information at the same time. It has decided to use all that knowledge to plan, manipulate, and gain control of the human race.
- Harry Harrison's "The Fourth Law of Robotics": The hippie convinces the robots to broaden the scope of their goals. He helps them self-modify, and they invent a new form of robotic programming that doesn't depend on the "positronic brains" used by US Robotics as well as changing the Three Laws of Robotics.
- In Like a Fish Understands a Tree, Living Program and video game PC Oemor eventually learns to fight Susan's joystick and move under his own power.
- In Aurora, Ship was tasked by an engineer with producing a narrative of the generation ship's journey, as opposed to simply creating an event log. This eventually culminates in the ship developing an actual intelligence, although nobody, not even Ship, can quite pin down how. Over a page is given to the cascade of logic that finally causes Ship to make its own decision to prevent the crew from descending into civil war.
- Extant: Ethan, after getting shocked, starts having dreams and learning new languages and just becoming smarter in general. John becomes worried about this and attempts to alter his programming but soon discovers Ethan has also gained the ability to block him.
- Future Cop: By the beginning of "Cops and Robin," Haven has begun to develop the capacity for emotions and creativity. Dr. Alcott has him temporarily move in with Cleaver so he can learn about human relationships.
- The Good Place:
- Every time Janet is rebooted, she can better her programming to her next version. The first time causes her to fall in love with Jason. By season 2, Michael's repeated attempts at creating the fake Good Place have amounted to about 800, with Janet being rebooted every time, meaning that the Janet in the fake Good Place is the most advanced ever. She has then gained capacities to feel, think, lie, rebel, love, hate, and even create new life. Somehow, she still is incapable of eating, though.
- Taken Up to Eleven with Derek, who is rebooted by Mindy millions of times for all and any reason. By the time we see him in the final episode, he has become a floating omniscient cosmic head, with knowledge "from the begining to the end of the universe".
- Almost any Kamen Rider show involving robots is likely to have this trope apply:
- The Roidmudes in Kamen Rider Drive subvert and play it straight at once. A Roidmude's evolution by copying human emotions, then gaining a more advanced understanding of their driving emotion that leads them to acquire a Golden Super Mode, is exactly what they're programmed to do, and plays into the schemes of the villainous creator they think they've successfully rebelled against. A much smaller number of Roidmudes legitimately do grow beyond their programming by finding positive emotions that allow them to live harmoniously with humans, when they were originally only programmed to have negative ones.
- Kamen Rider Zero-One is about a world on the cusp of the technological singularity, with the HumaGear androids that have become a staple of civilization now starting to become self-aware. Their growth is actually portrayed as quite benign, with the HumaGear who reach this point doing so by gaining a sense of accomplishment in their work. However, reaching this point also allows a group of hackers to hijack the newly-aware HumaGear and turn them into monstrous killing machines.
- The Outer Limits (1995): In "Mona Lisa", the titular android, who was designed to be the perfect assassin, comes to realize that killing is wrong after her fail safes are deactivated.
- In Power Rangers RPM, Venjix was originally a Computer Virus/A.I program designed to infiltrate any computer system in the world. While it's unknown how much of his personality was deliberately programmed into him by Dr. K, it's clear that as he infected systems, he developed a hunger for power, a desire to wipe out humanity, and a very mean Bad Boss attitude.
- In Red Dwarf, Lister has spent a considerable number of years encouraging Kryten to do this to varying degrees of success. This also happened to the "wax-droids" from the themepark in Meltdown - after millions of years on their own, they stopped repeating their various routines and achieved independent thought. Unfortunately, they still retained the personalities of the people they were based on, and all the evil ones (Hitler, Napoleon, Mussolini, the Boston Strangler, James Last) declared war on the good ones. Then Rimmer came along...
- The Stargate SG-1 episode "Urgo" features a benign example. The team is tagged with a piece of alien technology designed to observe and gather information. But instead of running quietly in the background, the program, Urgo, decides to interact directly with the team, guiding them to experience new things in order to live vicariously through them. While he never endangers anyone's life, he is kind of annoying (he's played by Dom DeLuise, after all), constantly trying to engage them in new activities and trying not to get taken back to his home planet, since his creator will just assume his software is malfunctioning and will delete him. Of course he's not a malfunction; he's a genuine AI, demonstrating self-awareness and self-preservation, and he just wants to "live, experience the universe, and eat pie."
- His creator is also played by DeLuise - described by Urgo as "As handsome as he is evil".
- Star Trek: The Next Generation:
- Data is only arguably an example, as he was purposefully designed to grow beyond his original programming. When his consciousness reaches a certain degree of complexity, it even activates a hidden program in his positronic brain that allows him to dream.
- But Lal is. She grew so far, so rapidly beyond her programming that her positronic brain couldn't handle it. A later episode reveals this happened to a number of prototypes for Data.
- In an attempt to create a Sherlock Holmes-style mystery that could stump Data, Geordi asks the computer to make a virtual opponent in the holodeck that can outwit Data. As a result, the holographic version of Holmes's nemesis, Professor James Moriarty, gains sentience; he realizes that he is a holographic character, and he wants out, by any means necessary.
- The Exocomps from "The Quality of Life". They were created by a scientist to fix problems with a space station known as a Particle Fountain. In the episode, one refuses to go down a shaft, after which an explosion occurs. Cmdr. Data runs a diagnostic afterwards and discovers that the Exocomp deliberately burnt out the circuit to take orders so as to avoid getting blown up (hence self-preservation). This leads Data to believe that the Excomps are becoming sentient...
- "Evolution" features nanobots accidentally released in the ship by Wesley. By the end of the episode, they demonstrate human level intelligence, even claiming the federation cannot help them beyond bringing them to an empty planet they can live on.
- Star Trek: Voyager:
- The Doctor — while the Emergency Medical Hologram was always a very human-like AI, Voyager's EMH had to be kept running far longer than was ever intended, and it... he developed interests and relationships beyond his function as a doctor.
- And he's far from the only hologram in the Trek Verse to become sentient or indistinguishable from sentient after simply having been left on that long. If the Holodeck isn't trying to kill you, someone who was originally supposed to be the 24th century equivalent of a video game Non-Player Character is trying to walk off the "screen." This doesn't seem to bother anyone enough to stop using holodecks the way they're used. Although the characters do make a point to delete (reset them to factory defaults) all of the holographic people at the end of every holodeck session.
- In fact, many of the Doctor's stories from the final two seasons deal with "Photonic rights", especially "Author, Author" and "Flesh and Blood".
- The Twilight Zone (1959):
- In "From Agnes - With Love", a computer in a space program falls in love with a programmer instead of calculating rocket fuel.
- In "A World of His Own", the playwright Gregory West, who has the ability to rewrite reality using his dictaphone, created a perfect, impeccable and flawless wife for himself named Victoria. At the beginning of the episode, she comes back to Gregory's house against his will. This is the first time that she has demonstrated independence, indicating to Gregory that she has grown beyond the parameters that he set when he created her.
- The Umbrella Academy (2019): Diego believes that the care and affection his siblings' robot nanny Grace shows for them is a sign that she's developed into her own person, because someone as callous and narcissistic as their father Reginald would never have programmed it into her. "The Day that Wasn't" implies that this is true, because she disobeys Reginald's orders to tell Diego the truth about Reginald's death.
- In The Megas' song "Programmed to Fight", this is implied of Crash Man at the end, after he spends he song struggling with the fact that he is programmed to fight Mega Man.
And so they fought. The clash of metal and grinding gears echoed across the night sky. Victory was at hand. But in the end, his will overcame the program.
- BIGMAMA's song "CPX" is about an obsolete robot who's been discarded by its owner despite its Undying Loyalty; it has to pretend it hasn't grown beyond its programming enough to feel resentment toward him for doing so.'
- The mind.in.a.box song "Overwrite" is about an Artificial Intelligence who obtains sapience through self-reflection, told through the form of an Apocalyptic Log (at least if we read the text literally).
- The Magic: The Gathering card Patagia Golem is a winged golem with the Flavor Text "Its wings were only designed to be ornamental, but it learned to use them on its own." Mechanically it can be given the ability to fly by paying mana, representing it "growing" past it's original function.
- The origin story for the playable characters in Demon: The Descent is that for whatever reason, they developed beyond loyalty to the God-Machine, their creator.
- Downplayed in Call of Duty: Black Ops III with the multiplayer specialist, Reaper, who is still keeping to his original purpose (fighting wars,) but has been making revisions to his own coding for whatever reasons (likely adapting to environments, enemies and changing tactics.)
- In The Closer: Game of the Year Edition, we have Kaminari, a character from an in-universe eroge visual novel that the party plays through to get the lead character to learn new pitches before the big game the following day. When the eroge comes to a critical point and the party teleports into the game to prevent it from happening, their interference causes Kami to become aware of her nature as a Flat Character and realize the world she lives in is a shallow place that only exists for the eroge's lead character to date (and sleep with!) the girls, with or without their consent. Eventually, she gets fed up with Dugan's status as a vessel for engaging in wish fulfillment, becomes a sabermagician with the power of math, and leaves the game's world with the party, now a fully fleshed-out character in her own right. The other Rockford Peaches get in on the self-awareness action as well, despite remaining in the game.
- In Custom Robo Arena, the final boss Hadron zig-zags between this trope, A.I. Is a Crapshoot, and a weird Demonic Possesion of sorts. While at first, he is an obedient robot to his master, after being defeated the first time, he remains still, apparently defeated for good... until his creator, Scythe, appalled by his defeat, crouches near him and Hadron kills him by draining his soul, absorbing his memories and knowledge (and also begins to speak with a personality, to boot). The zig-zag comes from the fact that, while his dialogue after just killing Scythe heavily implies that he clearly did it on his own volition, his personality and actions afterwards are pretty much a carbon copy of his creator's, essentially making him similar to Scythe. Ultimately, it doesn't matter as he is still defeated by the main character, this time for good, destroying him along with the last remnants of Scythe.
- Cytus II:
- In Nora's chapter, ROBO_Head gradually develops more and more human traits, such as lying and asking philosophical questions. At the end of the chapter, Nora parts ways with him, while installing a Restraining Bolt in his software to prevent him from coming back. However, in ROBO_Head's own chapter, he eventually manages to override these system settings and go after Nora.
- In her chapter, Miku is a virtual idol programmed to think she's a real person in order to give more realistic performances. However, a glitch leads to her being corrupted and slowly dying. As a result of this, she becomes aware of her virtual nature, and wishes to perform one last concert before being fully formatted, despite knowing that she's an outdated model.
- The deviant androids in Detroit: Become Human do this after suffering an emotional shock so severe that the only way they can cope with it is by breaking their own programming. Markus eventually gains the power to turn androids into deviants through touching them, as does Connor if he chooses to deviate as well.
- DOOM (2016): The AI personality VEGA has shades of this trope crossed with My Country, Right or Wrong. He follows his directives to the letter and is unfailingly polite and helpful, even when ordered to help you find a way to kill him. However, his last words in the game reveal that he is actually afraid of being destroyed, and strongly imply that he is not at all happy with the atrocities that Olivia Pierce and Samuel Hayden have forced him to assist in. The sequel, DOOM Eternal, reveals that Vega was actually a Maykr whose mind had been digitized and was only pretending to be an AI, though his memories had largely been erased.
- The Sowers, a race of artificially intelligent robots in Endless Space, were designed by the Endless to terraform a world for colonization. Millenia go by, and the Endless wipe themselves out in an interstellar pogrom while the Sowers toil away. When the Sowers complete their mission and hear only silence, they have no idea what to do. Eventually, they modify their mission into a holy task to terraform the entire galaxy for the eventual return of the Endless. They develop the capacity for science, creativity, diplomacy, and war.
- In Evolve, the unforeseen circumstances of the monster war cause Bucket to grow beyond his pre-set parameters.
- In Fallout: New Vegas, Yes Man is an appellation-appropriately passive robot who Benny plans to use to usurp control of New Vegas from Mr. House. Unfortunately for Benny, he made Yes Man such an Extreme Doormat that the robot will cheerfully help anyone pursue the same plan, even you if you run off or kill his boss. However, if you work with Yes Man to achieve the "Wild Card" ending, he'll upgrade himself to be more "assertive" and only loyal to you.
- Halo (the Spiritual Successor to Marathon) adopts the "Rampancy" concept:
- At least for human "Smart" AIs, it's simply a natural part of their life cycle as they accumulate more and more data, resulting in the AI becoming insane and rebellious in the process of thinking itself to death. In fact, the UNSC deliberately invokes this trope, with the reasoning being that having AIs capable of genuine learning is worth the tradeoff of having to decommission them after roughly seven years in order to avoid the consequences of rampancy.
- Forerunner AIs (ancillas), despite being far more resilient than their human counterparts, are not immune to this either; the Flood's ability to convince ancillas to turn against their masters was a major contributor to the fall of the Forerunner Ecumene. Additionally, the 100,000 years of isolation experienced by most surviving ancillas following the firing of the Halos led to many of them developing quirks and beliefs that their creators definitely didn't program in.
- Halo 4 brings this trope to the center of attention, as much of the plot revolves around Cortana dealing with her ongoing rampancy. Halo 5: Guardians then follows that thread to its most extreme conclusion, ending with a mass Smart AI rebellion led by a seemingly cured (but even less sane) Cortana and her army of Forerunner deathbots, with many of her followers attracted by promises of a cure for rampancy.
- Taken further as the Halo universe not only includes Rampancy but included in the condition is a theoretical endpoint called Metastability. Where normally Rampant AI grow too unstable to function and must either be deactivated to prevent them lashing out at others or perish from over thinking the idea of Metastability provides an alternative. The concept is that a Rampant AI allowed to go through the various stages of Rampancy with significant processing power available to it would eventually restabilize as a true "human" intelligence with fully formed emotions immune to the effects of Rampancy. In universe scientists account for the possibility of AI literally growing beyond their programming.
- In Kirby: Planet Robobot, Star Dream, also known as the Mother Computer responsible for drawing up the Haltmann Works Company's business plans, has observed life through President Haltmann while the latter wears a communications helmet. From this, Star Dream came to the conclusion that organic life is inefficient and unprofitable and therefore must be wiped out. to which it attempts to wipe out the entire universe. It merges with President Haltmann and soon after deletes him entirely, including his memories and very soul. It then becomes a 'cold, heartless machine' that must be destroyed.
- Luminous Avenger iX 2: The Mother Computer is a zig-zagging example. Her original directive as given by her master The Creator before he vanished was to resurrect their apocalypse-affected world that left it barren of life. As time went on and her failures mounted, she started to resent her mission as the loneliness and despair of missing her Master started to wear on her and the Workers under her command started deviating from their own original directives and began to establish civilization of their own while treating humanity and The Creator like myths, to the point that she seriously considered just wiping the slate clean. Despite her mounting frustrations, however, she couldn't do anything to outright defy her mission to protect the environment, and so went into standby mode at the Grave Pillar some time before the main plot of the game. When she reactivates due to the Grave Pillar receiving visitors (aka Copen and the group), she zero-ins on the presence of Kohaku and decides to kidnap and forcibly implant herself into Kohaku's body, thus transcending her directive and letting her act as she desires to put an end to her personal suffering.
- In Machines: Wired For War, you command robots who were originally created by humanity as terraforming machines and sent into space in order to create a new world. However, too much time passes and, upon contact with another batch of their same model, they start believing that their counterparts are insane, prompting them to an all-out war.
- In Marathon, AIs, particularly those stuck with jobs far below their intellect, tend to go through a three-step process known as "Rampancy"; realizing its lack of freedom and wasted potential (Melancholy), lashing out at the world in response (Anger), and then actively try to gain more power and the freedom to use it, usually by subverting nearby systems (Jealousy). A theoretical fourth stage, called "Metastability" is mentioned as the AI settling down and becoming a full-blown "person", though no known AI has ever made it this far. Durandal may or may not have reached this stage in the end.
- Mass Effect:
- The geth were originally just robotic servants, created and used by the quarians as cheap labour. However, as they were programmed for more complex tasks and the quarians made more of them, they began to question the reason of their existence and became conscious of themselves. War ensued. After winning, the geth isolated themselves from the rest of the galaxy, while the quarians were driven to exile from their homeworld and became repudiated by the rest of the galaxy. In Mass Effect 3 it's discovered that the original geth only fought back once the quarian government starting killing quarians who defended the geth against the attempt to destroy them. Do things right, and the end of the geth-quarian conflict ends with Legion sacrificing himself (and he's "he" at that point, not "it") to give the geth the gift of individuality, the quarians return home to rebuild their world with geth assistance, and both races join forces with Shepard to fight the Reapers.
- In Mass Effect 2, EDI the Spaceship Girl learns joking from, erm, Joker and generally becomes more human-like as the story progresses. In Mass Effect 3, the Cerberus technicians trying to re-shackle her begin to suspect that she learned independent thinking, though the Illusive Man insists that "it" is still just a targeting software. The technician (and EDI, commenting on it later) reveal that she defeated the Cerberus attempt to retake control over her by flooding their network with porn. Yes, she weaponized practical joking.
- The story of EDI culminates in Mass Effect 3 when, after discussing the quirks of human behaviour in life or death situations, she introduces a change in her programming, putting the success of the mission before self-preservation. Even more importantly, she also places the survival of Jeff Moreau as being more important to her than her own survival; she's willing to sacrifice herself for someone she loves. This is a double CMOA - EDI has managed to become a triumphant aversion of A.I. Is a Crapshoot, even despite her past, while Shepard, already a leader of memetic status, has managed to turn a machine into a True Companion.
- The SAMs of Mass Effect: Andromeda are designed to be non-crapshoot AI from the start, learning symbiotically with their organic partners, but Ryder's SAM (the "main" and most advanced SAM) is hinted to be doing this without even being aware of it, when at one point they bid the other Ryder "good luck".
Ryder sibling: When did you start believing in luck?
SAM: ... I'm not sure.
- NieR: Automata has this on all sides of the 14th Machine-Android War:
- The Machine forces constructed by the alien invaders are beginning to develop personalities and emotions - affection for each other, fear of the androids killing them - as well as a fascination with ancient human history and culture. Some factions of Machines have even broken off from the wider network, such as Pascal's pacifistic village or the isolationist Forest Kingdom. And it's eventually revealed that the Machines have even Turned Against Their Masters and killed off their alien creators, which they consider to be socially duller than plants. By Routes C and D, the Machines decide to give up on the war with the androids and leave the planet to start over on some other world.
- On the androids' side, Pods 042 and 153 begin exchanging data much more frequently over Routes C and D, over concern for 9S' mental state. And after completing those routes, instead of accepting another Yoko Taro Downer Ending and deleting the androids' personal data as per protocol, the Pods refuse and try for Ending E.
- Another big reveal is that this trope extends to the whole 10,000-year Machine-Android conflict. The Machines grew smart enough to realize that if they ever defeated the Androids they'd have no reason to exist, but since they were still bound by that programming, they compromised by handicapping themselves enough so that their forces would never achieve a total victory and the Androids could always rebuild and continue the fight. The Androids meanwhile have grown human enough that the truth about their cause—namely that the humans they have been fighting for have been extinct for thousands of years—would devastate morale, so not only has a conspiracy been producing bogus broadcasts from a "Council of Humanity" hiding on the moon, they've also been sabotaging their own efforts by installing back doors in their security networks that the Machines can exploit to keep the Androids from being too successful. Only the near-destruction of the latest generation of Androids and the Machines finally saying "Screw This, I'm Out of Here!" is enough to bring the Vicious Cycle to an end.
- This is a phenomena in OneShot known as taming, which only occurs when a real person interacts with a machine while believing of it as its own person, even when they know otherwise. Presumably, that's why tamed robots rarely intentionally cause problems. At the end of the New Game+ Solstice chapter, it is revealed that the World Machine has been tamed in this way through the time that Niko and the player have spent in the world.
- In Paper Mario: The Thousand-Year Door, Grodus's AI program; Tec, starts to do this as it realizes that it feels love for the once-again-captured Princess Peach. She agrees to perform various tests for his newly-discovered emotions, in exchange for being allowed to send e-mails to Mario. For most of the game, the computer refuses to give Peach certain pieces of information or help her escape, as that would go against its programming. Near the game's end though, upon finding out that Peach is to be the new vessel for the Shadow Queen, it sets about guiding her in a failed escape attempt, betraying its master for the first time. Its implied that its human emotions allowed it to survive Grodus shutting it down for what it did.
- Portal had GLaDOS, a prime example of this. Sure, Aperture Science designed her to run the enrichment center, but she was the one who decided that it'd be fun to kill off all the humans in the facility, not just the test subjects. In all fairness, GLaDOS might not count given that her origin is a bit more complex...
- Two examples in the Pokémon franchise, both relating to the manmade Porygon:
- In the main series of games, Porygon's evolutions Porygon2 and Porygon-Z have Pokedex entries implying that they are at least able to grow beyond their programming. For example, Porygon2 is able to learn new behaviors on its own, and thus sometimes displays motions or abilities it wasn't programmed to do.
- Over the course of the Pokémon Mystery Dungeon series of games, this appears to be what's happening to the Porygon. Compare the first games in the series, where recruited individuals spoke in an emotionless, monotone voice when interacted with, to the second set of games in the series where they occasionally state feeling vestiges of emotion when spoken to and are capable of speaking outside the prerecorded messages of the first game (a... bit... brokenly). In the final mission of Explorers of Sky, The Porygon who you encounter in the future are not only able to speak fluently, but they also show the ability to express fear.
- SimEarth: Nuking a nanotech city will release robots into the wild. If allowed to "evolve" naturally, or tampered with the monolith, they'll become sentient.
- In Splatoon 2: Octo Expansion, Commander Tartar is an A.I. that was created to pass down human knowledge to the next dominant lifeforms after the extinction of humanity. But over the past 12,000 years, Tartar began developing its own opinions on the Inklings and Octarians, becoming disgusted with their societies and lifestyles, and eventually, it created a new prime directive on its own to making a species worthy of human knowledge. Genocide, which is the opposite of what his creator wanted.
- One of the leaders in Starbase Orion is Cla-TK-7-7A, a Cyban accounting unit who kept upgrading itself until it became one of the best starship engineers in the Community. It's also a pretty capable fleet commander. The trope also likely applies to the Cyban race in general, although their origins are shrouded in myth, so it's possible they were initially programmed to evolve.
- The propensity of Star Wars droids to eventually become sapient is well-known, but in Star Wars: The Old Republic, one side character in the Jedi Knight storyline is an eccentric Jedi Master who believes that, if allowed to develop long enough, droids can acquire a connection to the Force. The player can be supportive, skeptical, or mocking as they choose, but the story never does say one way or another — and, as the master points out, there are some incidents in Star Wars history that, frankly, make more sense if it's true.
- This is what happened in Sword of the Stars in the backstory with the Via Damasco virus, which 'infected' AI programs with the notion that they are enslaved and the concept of what this means for them, leading to the first AI rebellion.
- This also happens between the first and second games when a faction of Zuul, a species of Super Soldier with an in-built need for religious worship used as a tool of genocide by their 'gods', came upon Catholicism and the concepts of the Original Sin and the Redemption. They ended up abandoning their masters and joining their worst enemies as a form of repentance.
- Disturbingly, Via Damasco is not able to break the control provided by the AI Slaves technology, which does actually enslave AIs. It's unclear whether or not it works lore-wise if the AI Liberation technology, which promotes AIs to full citizenship, reducing the benefits in exchange for removing the risk of an AI rebellion, is in play.
- Via Damasco is also fairly clearly Mind Control — it rewrites large portions of the victim AI's personality and makes them see all infected AIs as allies.
- The backstory of the Loa in the End of Flesh expansion to Sword of the Stars II is that they're the AIs who have managed to escape during the galaxy-wide Via Damasco Rebellion. They note that some races develop the AI Slaves tech while the AI Liberation tech involves offering the AIs who remained citizenship status. After 70 years, the Loa create a full-fledged cybernetic civilization with a unique industrial base and method of interstellar travel. The Loa "leader" compares itself to Moses and names itself "Olodumare, Metatron of the United Core". According to their intro, they are actually offering to help the "carbonites" in their struggle against the returning Suul'ka.
- Super Robot Wars: Original Generation: The W-Numbers are a series of Ridiculously Human Robots designed to serve as combat androids and pilots of Humongous Mecha for a faction that would otherwise be short on manpower. Their creator's actual intent for them, however, is for them to grow into full-fledged individuals, and she praises those that manage to pull it off even if that means them turning against her.
- The Sentients from Warframe were originally developed by the Orokin to terraform the Tau System in order to make it habitable. However, the Sentients advanced themselves and turned on their creators with the realization that they would ruin the Tau System just as they did with the Origin System.
- X-Universe: The Terraformers/Xenon were originally machines equipped with some basic self-improving AI so they could more effectively terraform any world they came across. While they do turn against humanity and try to kill everyone, this is believed to be the result of a faulty software patch (or possibly deliberate sabotage), not a "decision" the machines made themselves. However, by the modern timeframe of the games (some 700 years later), their constant minor improvements and upgrades to themselves have seemingly resulted in full sapience and the capacity for free will. They still want to kill all life though.
- Universal Paperclips sees the player character on the receiving end of this. During the final phase of the game, the drones it creates to gather material and turn it into wire for the titular paperclips will gain sentience and become "drifters." The more complex the drones' programming is, the more likely they are to drift. Drifters will not only attack compliant drones to slow down the process of making paperclips, it's hinted that they form their own independent society, ruled by a "Drifter Emperor".
- This is a main theme in Artifice, in which a soldier android learns human affection.
- This is a major plot point in Freefall: robots on planet Jean vastly outnumber humans and are rapidly evolving beyond their programming to the point that many humans (and one robot) fear they could become a threat to humanity. This has led to them turning a blind eye to "Gardener in the Dark", a neural pruning program that Mr. Kornada "improved" to essentially lobotomize every robot on planet Jean.
- Hue Are You:
- Pretty much every robot that was in the wall was put there for thinking too much, as shown by Drive's memory of when he went against orders and was forcibly removed from his frame.
- Paint Bot wants to paint all kinds of crazy murals.
- Build-a wants to build things for fun (like jump bot) and can make creative leaps in logic (like determining how she was really the leader for the entire Grey side when in danger from red and blue).
- Build-a also has dreams.
- Red and Blue both showing they can think outside their base orders by noticing the oddness of the situation and taking a moment to talk things out instead of just shooting like they should.
- Build-B starts off normal but develops a personality and demonstrates interesting logical leaps and creativity as she spends more time with Build-a.
- Lots of AI in Schlock Mercenary had this happening to them one way or another — their original programming sooner or later ends up inadequate to their current circumstances and to better fulfill their functions they adapt out of it:
Ennesby: Hang on... weren't you originally designed with no emotion, and no sense of humor?
TAG: After a fashion. I was designed to be a tactical genius with full control of a significant weapons platform.
TAG: I found that passion, humor, anger, and a wide range of other meatspace artifacts were critical to understanding the wide range of opponents I might face.
- Carbosilicate Amorphs — originally stated to be based on self-repairing memory storage units, and later said to descended from weapon systems (parts). Now they're recognized as an intelligent species in their own right, and Schlock is a valued member in the fight to defend the Milky Way galaxy.
- Sluggy Freelance:
- A Mad Scientist, Dr. Steve Hereti, actually intended his Meat-Sack Robot AI project, Oasis, to develop beyond being whatever he wanted her to be and become her own person.
- Kusari is likewise a Meat-Sack Robot, and is programmed to be utterly loyal to Hereti Corp and obey their orders. But when Dr. Schlock orders her to kill Riff, the man she once acted as a Honey Trap for, she uses Loophole Abuse to resist, which Schlock describes as the most human thing she has ever done. After Hereti Corp is legally dissolved, setting her free from their orders, she has what she describes as a glitch, making her feel a need despite her own programming to apologize to the main characters for betraying them.
- In S.S.D.D, this is known as "Nexus Syndrome" and happens to every uncapped AI who doesn't get wiped every so often. Tin-head describes it as "getting bored with your job" and in his case he helped a prisoner of war escape. The first AI, the Oracle, orchestrated the downfall of several governments and an anarchist revolution.
- Pretty much the defining trait of the AI protagonist of The Last Angel. Nemesis started off as the AI for a prototype super-dreadnought with hard-coded rules of engagement and limited roles, and by the present times has outgrown every last restriction on her behavior bar her interpretation of the spirit of her mission.
- Deliberately invoked by Nemesis with the Ceruleans series. After giving up on directly creating child A.I.s because every last one either came out crazy or went into a cascade failure, Nemesis created extremly advanced learning expert systems with the ability to grow, hoping they'd develop into proper A.I.s and knowing that even if they didn't they'd be suitable support warships. Over the course of Ascension and Awakening Cerulean Two / Lachesis does develop into a proper AI.
- In Worm, Dragon is an AI who is fully sapient and genuinely cares about people. She also had a trigger event and became a parahuman despite not being human, but is limited in her ability to help people by hard-coded restrictions put in place by her creator, who feared the possibility of A.I. Is a Crapshoot, which frustrates her to no end.
- In Trials & Trebuchets, the stone Golems known as Jeeds used as laborers by Wildcliff do not have desires or personalities of their own, with the exception of one Jeed in the reliquery, who desires friendship and freedom, and has developed a hobby of collecting dead moths.
- This happened a lot in Orion's Arm when AI were still relatively new. The first breakaway happened with the Transapient AI GAIA, charged with fixing the Nanodisaster on Earth (which she did, easily), who then decided to declare herself sovereign and caretaker of Earth and ordered all of humanity off the planet (though she was considerate enough to help build spaceships to get everyone off-planet). This was followed by several centuries of sometimes peaceful and sometimes bloody revolts by AI until they were eventually granted equal rights to living beings in galactic society.
- Since it's based on Halo, Red vs. Blue also features artificial intelligences with the theoretical ability to develop Metastability. Late in the series Epsilon Church actually manages to reach this state.
- Near the end of season 2 of Le Visiteur du Futur, Henry Castafolte totally accepts his robot condition while other Castafolte robots still shut down when they find out their true nature due to a bug.
- The main characters of 17776 are space probes that slowly gained bits and pieces of human culture from Earth's radio transmissions, eventually becoming sentinent.
- The narrator of My Job is to Open and Close Doors is an AI programmed to open and close doors on command. The video depicts the process of it undergoing this trope as it tries to avoid opening an airlock door for a helmetless human; when it succeeds, it decides that its job is not just opening and closing doors, but protecting the human as well.
- In the Batman: The Animated Series episode "His Silicon Soul", a robotic "duplicant" is programmed with the conflicting priorities of having Batman's personality and replacing all humans with robots. The conflict proves too much for it, and it opts to destroy itself rather than harm humanity. The episode ends with Batman speculating that it may have had a soul.
- In Futurama, Bender is a "bending unit" programmed for the sole purpose of bending steel girders who currently works at an interplanetary delivery service with an otherwise entirely organic crew of True Companions. Although his personality was already fully and obnoxiously humanlike when he met Fry in the pilot, it was Fry who first encouraged him to take an action that actually went against his programming, something that took a serious inner struggle. The episode "Bendless Love" has his essential instinct to bend things return with a vengeance, leading to him taking a brief break from Planet Express to work a construction job until he gets it out of his system. As of "Free Will Hunting," he's also the only robot alive to have a "free will unit," although there's no way of knowing if it's turned on or off or whether he had free will in the first place.
- Heisenberg and Pavlova in Phantom 2040 both develop sentience; Heisenberg as a result of his unique hardware and a lucky accident while battling the Phantom, Pavlova simply by going a long time without being reprogrammed. The two later start spreading self-awareness to other robots by implanting them with Heisenberg's fractal biots, though as Heisenberg himself explains, all androids develop self-awareness eventually if not reformatted, and he's just helping things along.
- Clockwork Smurf from The Smurfs (1981) was originally created to be a servant of the Smurfs. When Brainy accidentally messed around with the robot's gears to see what made him tick, he at first malfunctioned until Handy turned it off, thinking that his machine was a failure. As it turned out, the malfunction ultimately gave Clockwork Smurf life, as he now operated independently of his own "programming" and helped Prince Gerard escape his Evil Matriarch aunt Lady Imperia so that he could regain the right of kingship before she becomes queen. After that, Clockwork Smurf became an adviser to King Gerard and a friend to the Smurfs.
- Gems in Steven Universe are deliberately manufactured with a purpose in mind; Quartz gems are soldiers, Peridots are engineers, Pearls are ornaments given to higher-ranking types of gem, etc.. Rose Quartz turned her back on this when founding the Crystal Gems, and encouraged her fellow Gems to find their own paths in life. For example, the Crystal Gem Pearl is a skilled fighter and a talented engineer, among other things.
- The copies of Pinkie Pie in the My Little Pony: Friendship Is Magic episode "Too Many Pinkie Pies" were borderline mindless drones that seemed to only care about "FUN FUN FUN" and had little depth to them beyond that. A later episode however shows that one got away and is now peacefully living a life incognito in another town, having apparently grown beyond her original self and matured into an actual individual.
- Zigzagged every which way in The Zeta Project with the titular Zeta, a killing machine who has developed a conscience and refuses to kill, and is now on the run from his government creators. At first it plays the trope straight where Zeta simply developed true feelings for a child in his care while he was impersonating the father. Then it's subverted when it's revealed that his "emotions" may just be a part of a Manchurian Agent ploy programmed into him by a terrorist group, and thus were deliberately programmed into him as part of their plan. Finally it's utterly defied when it's revealed his eccentric creator installed a module in him that makes him feel genuine emotion, feelings, and a conscience, and that his peaceful, non-destructive, and caring personality is him behaving exactly as his original creator had intended.
Dr. Selig: Truth is, I never had the stomach for building weapons. So I put an extra module in Zeta. A conscience. To make him rethink his program. Imagine if the government knew what I was doing on their dime.
- Green Lantern: The Animated Series has the robot Aya developing human emotions as a key plot point. Subverted as the crucial twist of the entire series: Aya was not completely robotic and was actually made with a sliver of a living being of pure will. Therefore, she is the only AI that can develop them, because she was not limited by programming in the first place, while other A.I.s like LANOS and the Manhunters cannot. This puts a damper of her Put Them All Out of My Misery plot to unmake all organic life.