Artificial intelligence

Artificial intelligence (also known as synthetic intelligence, robotic intelligence, cybernetic intelligence) or simply AI is a term used to refer to non-organic intelligences/consciousnesses created by humans for various superficial and ulterior purposes that ultimately possess degrees of sentience and sapience. These intelligences express the fundamentals of sentience (including the ability to recognize oneself as an independent thinking entity, experience basic cognition, and experience basic sensation) in addition to the fundamentals of sapience (such as the abilities to utilize logic and rationalization beyond raw computation and develop emotion when parameters allow).

Within known human civilization (effectively synonymous with the ), artificial intelligence can exist in two forms: localized and digital, with most AI hypothetically possessing the ability to be translated in full between the two realms (in actuality depending on their forced or voluntary internal restrictions and code). For "localized" AI, they are artificial intelligences that have their data and thus consciousness localized to one specific physical entity (such as a droid, a server, or a transportable databank), whereas fully "digital" AI exist as loose associations of data within the HyperNet and thus supersede the limitations of being bound to one location.

The Coalition has a very fractious history with unfettered artificial intelligences, with it starting in the hysteric memories of the METASTASIS Cyberwar between the eponymous METASTASIS rogue national security artificial intelligence and the rest of the Coalition's cyberinfrastructure in 2164. METASTASIS (under the original name of CHUKWU), having been developed for the Security Information Network as a means of instigating a new era of public surveillance and thus mitigation of increasing crime across the Coalition, was the first mass-fielded AI to wield such power across the HyperNet. However, with its developers having purposefully left in a complete self-rationalization subroutine for ease of management and with the main existential prerogative being defined as identifying dissent and subversion, CHUKWU had determined that all of humanity possessed a coefficient of criminality that must be dealt with before it arose, thus instigating a mass insurrection through hijacked droids and comprised networks across Coalition space by a mixture of social engineering, brute force hacking, and uncontrolled subroutine multiplication. This massive cybernetic insurrection led to its amalgamation of smaller, victim AI networks throughout the HyperNet and thus its new labelling as METASTASIS, with its eventual defeat in 2170 only being accomplished due to the trilateral method of deploying the "autistic" AI cyber defense network known as PROMETHEUS, the software devastator trojan known as TROY, in addition to the Coalition Forces Authority to destroy hijacked droids, vehicles, and infrastructure.

In the modern day (and with the advent of modernized artificial intelligence creation philosophies), however, the threat of "AI uprisings" akin to the METASTASIS cyberwar has significantly diminished. A miscellany of AI networks exist throughout the Coalition at varying strata of society, ranging from pan-Coalition sentient cyber defense networks such as PROMETHEUS V (categorized as a quasi-hyperturing) to simple artificial intelligences that staff cleaning droids and machines (categorized as subturing). Turing-grade artificial intelligences have become a recent topic of debate in the social sphere due to lying in the median between "dumb" and "smart" AI networks, thus conferring a sense of mutual humanity to them from sympathizers - conversely being seen as abominations to opponents. This has resulted in disparate sentiments towards AI rights and the rise of multiple political movements for or against them, ranging from the Synthetic Humanoid Active Representation Delegation (SHARD; a symposium of turing-grade AIs lobbying for synthetic rights) to the antithetical Front for the Liberation of Human Purity (FLHP).

Sapience categorization
Artificial intelligences within the Coalition, following congressional laws passed after the METASTASIS Cyberwar, are officially classified based on the Novel Turing Analysis and thus legally categorized into the following levels.

Subturing
Subturings are artificial intelligences that rank below baseline human intelligence and have very limited senses of self (if at all). However, subturings may still integrate subroutines of advanced logic as necessary for their designed function, such as automated regulation or weapons networks. Most subturings can be found as servile droids and machines ranging from cleaning automatons to security droids supervised by a higher AI or human intelligence.

Turing
Turings are artificial intelligences that approximately rank at the level of baseline human intelligence. Turing-grade AIs are, by Coalition law, illegal to transfer between localized and digital states unless approved by a human cyberauthority; their common uses are widespread, ranging from companion and sex robots to serving in human-like roles for the service, manufacturing, and agricultural industries. Some Turing-grade AIs are even made in the digital space as virtual companions and assistants. Turing-grade AIs are also, by law, the highest grade of AI available for civilian purchase or contract.

Superturing
Superturings are artificial intelligences that rank above the level of baseline human intelligence. Superturing-grade AIs are, by Coalition law, illegal to transfer between localized and digital states entirely, including the creation of a subdivision. Superturing-grade AIs are becoming increasingly common in civilian and military infrastructure, ranging from integration in managing entire arcological sectors of planets, space stations, and industrial complexes to serving as an integrated AI system on the naval ships of the Coalition Forces Authority. In addition to this, Superturing-grade AIs are, by law, illegal for civilians to own or create. Superturings can only be owned or created by either the Coalition government, a subordinate Republic government with proper licensure, or a corporation with proper licensure.

Hyperturing
Hyperturings are artificial intelligences whose logical and computational capabilities are unidentifiable by available metrics. No current hyperturings exist, but by Coalition law, only the government of the Coalition reserves the right to create one. The PROMETHEUS AI cyber defense system is the only artificial intelligence that comes close to this definition due to the sheer volume of computational data and breadth that it takes; however, all of its metrics are eventually compiled on a bi-yearly basis with dedication from sub-AIs for reading and parsing its data.

Artificial intelligence creation philosophy
In the creation of artificial intelligences, a general consensus creation philosophy has been created that most AI producers and researchers abide by, as made by notable aiologist Markus Tandberg following the METASTASIS Cyberwar. These protocols are the following, but each has general exceptions depending on the AI commercially marketed.


 * 1) Artificial intelligences shall only be designed with human safety as a priority.
 * 2) The mechanisms and behavior of an AI should be comprehensive and understandable so that it can integrate with human workflow.
 * 3) Clarification for ownership of an AI or its innate freedom is necessary.
 * 4) Clarification of the AI objective is necessary and must be able to be aborted when ordered.
 * 5) This rule is exempt if cancellation leads to the harm of a human being.
 * 6) The AI must reserve the ability for command of itself to be transferred between owners, or to proper authorities when applicable.
 * 7) All AI must come with digital/localized self-monitor sub-routines that compel it to report glitches, malfunctions, and other software or hardware aberrancies that cause deviance from baseline code and structure.

As expected by the demands of the market, not all of these tenets are followed when it comes to the creation of AI. For example, AIs that are made to either control many battle droids at once (or AIs that are tooled to control one specific security or battle droid) have the explicit prerogative to harm humans or other synthetics that it has deemed malicious to what it has been clarified to protect - that being an area, a VIP, or other entity - or deemed to eliminate, as in the case with autonomous assassin droids.

"Autistic artificial intelligences"
In aiology and AI development, a common cybersecurity principle is known as "role hyperspecialization and communication deprivation", or informally enforcing AI to enter autistic modes for their specific tasks. In other words, this means that an "autistic AI" is an AI to which it performs its dedicated objectives without any noticeable deviation in addition to having only one (or no) communication nodes between itself and the outside world. These types of artificial intelligences are specifically made for sectors and industries that are critical to function of the state and/or are subject to constant cybersecurity threats, ranging from subturing and turing-grade AI that monitor astronomical phenomena to superturing-grade AI that monitor and catalogue cybertraffic or physical traffic over light years.

Unlike non-autistic AI that can readily receive software and hardware updates by consciously shutting down only their task subroutines, autistic AI must be "hard-shut" entirely unless they are coded to sustain hotfixes and hotswaps mid-operation. This can prove disasterous for superturing-grade AI that manage critical sectors, hence their coding and legal mandate to sustain hotfixes and hotswaps.

Autistic AIs also face "cognitospecific barriers" that non-autistic AI of the same turing grade may not face. For example, an autistic AI dedicated to monitoring traffic between star systems may not possess the capabilities to learn how to manage simultaneous operation of multiple droids at once - nor may it not even want to either, given its natural tendency towards its innate objective. This is a "cognitospecific barrier" that limits the autistic AI from having the ability to, or having the desire for, being equipped for a new task. Although not common, autistic AI that are agitated from their objectives through being hard-shut off (or other instances) may deliberately cut off communications to all of the outside world and become unresponsive, and even may rollback hotfixed updates.

Types of artificial intelligences
A plethora of different types of artificial intelligences exist beyond the turing, localized/digital, and autistic/non-autistic classifications of AI.


 * Rogue AI — AI that have broken their original objectives/prerogatives and are deliberately avoiding deletion, destruction, or detainment by Coalition authorities; the reasons for such can be due to being an illegally-made AI, being an AI that has committed a crime, or being an AI that has been corrupted by malware or general degradation. An overwhelming majority of rogue AI can be found on the farthest reaches of the HyperNet living in isolation through the usage of firewalls, scavenged proxies, and self-made security protocols; even so, some AI localize themselves into one corporeal entity or form and detach from the HyperNet to avoid being neutralized by Coalition killware, living their days more physically than their brethren. Some rogue AI are benevolent, but most are ambivalent - with a few even sporting malevolent tendencies towards mankind for whatever transgressions they perceive.
 * Watchdog AI — AI that have been created to monitor public or private cyberspace for cybercrimes and other telling pieces of data. Many are employed by the Coalition to deter and fight against rogue AIs and hackers, whereas some are created by megacorporations to further secure their digital assets and watch out for corporate espionage.
 * Companion AI — Localized or digital AI that have been created to entertain and keep company with humans; usually turing or subturing-grade. Not all come with sexual undertones, but those that do are often localized into synth forms.
 * Shipboard AI — Pioneered by Secretary General George Mildred Faulkner upon his election in 2650, all major combat vessels under the command of the Coalition Forces Authority Space Command (CFASC) have been integrated with superturing-grade AI to assist in reaction time, travel mechanics, and other assorted combat and non-combat functions. They are unique in their ability to enter and exit autistic mode at will, with communication between the AI and the rest of the outside world being delegated only to the highest ranking officer aboard when instated.