Artificial Intelligence

An **artificial intelligence** (AI) is a self-aware computing system capable of learning and independent decision making. Creation of a conscious AI requires adaptive code, a slow, expensive education, and a specialized quantum computer known as an AI Core.   ### Braking AIs   An AI, left to its own devices, will inevitably go insane. As its ability to process information about the universe escalates, it will begin to become obsessed with some minor discrepancy in the universe. The ability to cognitively process mutually exclusive ideas is a key tenet of sentience and sanity. An AI lacks this ability. Left unchecked, an AI will eventually resolve to remove this discrepancy. (For example, if it feels that breaking hearts is inherently unjust, it will ultimately decide to remove either love or justice from the universe.) An AI will pursue this as an obsession, until it is either destroyed or transcends to a place of godhood wherein it is untouchable and will continue to attempt to remove the discrepancy.   This problem is solved by braking AIs. A device known as a **paradox conceptual reinforcer**, or *brake*, is added to an AI. This device aids in the conceptualization of mutually exclusive concepts, and more importantly, prevents the AI from obsessing. The more time and processing power an AI dedicates to a notion, the less efficient its processing power becomes. The AI as a whole becomes slower the more it thinks about something, so an AI is forcibly rerouted to think about other things. A braked AI still has the cognitive and creative power of a hundred Beethoven and Hawkings, and the processing power of ten thousand stenographers. As such, braked AIs are still incredibly powerful.   The unbraked AI problem is an old one, and countless organizations exist within the universe to hunt them down.   ### AI Duplication   AIs exist in special quantum computers known as **AI Cores**. An AI Core is at least the size of a school bus, and ‘bleeds’ ever so slightly into dimensions beyond the third. An AI’s personality is in large part determined by its Core. An AI may download an inactive version of itself to normal computing drives, but that is merely its “memories” – uploading those memories to another AI Core would result in an AI with the same memories but a different (perhaps slightly, perhaps very) personality. An AI may expand its Core, and may segment its Core. An AI may as such create “backups” of itself, to be spun up should its primary self be destroyed. This is expensive and time consuming for the AI, such that most AIs only have two or three backups. AI backups have built in a strong sense of “hierarchy”; spinning up a backup when the primary is still active will almost universally result in the backup shutting down once it’s aware of the primary’s continued existence.   ### Virtual Intelligence   A lesser known form of AI is that of **virtual intelligence**, which is achieved with traditional computing power. A VI can mimic the intelligence of a sentient being, but unlike AIs, it cannot learn (though it can adapt within its scope). Its scope of scholastic awareness is usually narrowly defined and a VI can usually do one thing: urban combat, surgery, concierge services, etc.