"Three Laws"-Compliant: Difference between revisions

Line 193:

** Also [[Averted Trope]] in cybercrime and cyberwarfare.

** [[Attack Drone|Predator drones]] are decidedly not first-law compliant. This may be an aversion, though, since all weaponized drones have operators who control said weapons most of the time, and are at least monitoring everything while it's active. The drones are usually only automatic when they're flying around and taking pictures. They're currently more remote pilot than AI pilot.

** In [[The New Twenties]] there have been experiments made with AI fighter jets, with the [[Yanks With Tanks|U.S. Air Force]] conducting a simulated dogfight between a human pilot and an AI piloted craft in 2024 according to an [https://www.theverge.com/2024/4/18/24133870/us-air-force-ai-dogfight-test-x-62a article] by ''The Verge''.

* Often averted in practice. It relies entirely on the programmer to enforce this trope in real life. There is no FirstLaw() unless it is written. Compliance is only as good as programming and design.

** For robots that work in a cage, the assumption is often made that humans will not be in the area while the robot is on. If humans do enter, there may not be any system to detect and work around them, resulting in potential harm. There is a reason that it took some time for robots capable of sharing a work environment with humans to take off.

** For robots that interact in the same spaces of humans, especially where humans aren’t expecting robots, interactions easily cause injury or death through bumbling or unexpected behavior, weapons or malicious programmer intent not required. Even with [[Genius Programming]], one can’t account for everything.