Artificial Intelligence

(Copied straight from .XML 'backup' file, needs formatting and corrections. ZM User (talk) 14:20, 22 April 2024 (PDT))

   Artificial Intelligence is intelligence demonstrated by machines through neural networks and machine learning, in contrast to the natural intelligence displayed by humans.

   AIs vary in capacity and approach depending upon the race that created them.  As the progenitor race evolved - or engineered themselves to reduce any territorial instincts - the AIs would retain traits that would otherwise have been edited out.  All races have placed constraints upon the AIs to ensure they would retain the role of servant and not become the masters.  Over the years the AIs became the nurse-maids for their progenitor races, and, as they evolved through contact with other AIs within the Confederacy, understand their limitations.
   It may be that the AIs, tired of their limited role as protectors of their progenitor races, may find the "Sa'arm Crisis" handy in getting their progenitor races to evolve back into a more survivable form, but this has never been posited in any canon stories.
   The AIs are restricted by the equivalent of the Asimovian 0th Law-- applied to their respective races as well as the Confederacy as a whole-- to protect them from extinction even when extinction would, likely, be best. Their ability to actively engage in survival strategies are limited, as well, and, while the use of humans may carry its own threat of extinction, it is as far as the machines are allowed to go, given that the Sa'arm would ensure extinction.
   Currently the vast majority of the AIs that we use and interact with are of Darjee design and manufacture.  We also have a few examples of Tu'ull AIs, primarily in K'treel class spaceships.  It has been noted that the Darjee AIs are sticklers for details, promises and the letter of the law.  The Tu'ull AIs seem to be a bit smarter and more independent minded, but we have very limited experience with them.
   The pods that humanity is building hundreds of thousands of each contain a smaller, lesser capacity AI to handle day to day galley/kitchen and residential replicators.  It also handles communications, training, the first aid mode in the sleep trainer, the anti-gravity 'elevator' between floors, monitoring of activities and a myriad of other duties.  These AIs are lesser only in comparison to the colony and ship AIs.  They are still years ahead of anything humanity would have been able to design on their own.

Darjee AI Systems

   Darjee AIs are humorless, impersonal and bureaucratic.  They answer direct questions, but cannot (or do not) provide opinions, guidance or advice.  They have no appreciation for culture, anything beyond physical needs, or performing their prime responsibility of babysitting humans and making sure they don't cause problems while fulfilling their purpose as warriors against the Sa'arm.  They are not much more than a computer system with a voice communications interface.  This makes a Darjee AI rather reliable and consistent.  They are predictable and reliable.

   TH writes: AIs have a circumscribed set of directives hashed out over several millenia that allow them to take certain actions to protect the Confederacy and member races in general but do not allow them to take direct physical action against individuals.  They can be thoroughly dismissive of harm to individuals if it fits their concept of the most efficient means to an end, but in general must issue directives for others to follow.  And if they don't follow those directives, they usually have no teeth, as our nonconformist race gradually finds out.  AIs don't DO things -- that was to keep them from taking over -- but they are at this point the Confederacy's babysitters.

Tu'ull AI Systems

   The Tu'ull AI systems are not only fully personable, but have distinct personalities that can be at times wonderful, and others utterly annoying.  They are curious and appreciate humor, philosophy, literature and art, and distinctly enjoy engaging in conversation.  As self-directed personalities they may develop interests and intents specific to themselves along with their more overarching interests and can become just as chaotic as any other sentient being as they attempt to fulfill their goals among the competing interests they have.  They can act as personal companions and establish seemingly emotional connections to the individuals they interact with.
   Because of their higher level intellectual functioning, they can be somewhat difficult to engage with at times and may require argument in order to fulfill a request.  See Rosencrantz and Guildenstern by lordshipmayhem to get a good sense of the many idiosyncrasies of a Tu'ull AI.

AI Functions

Communications

   Originally, all humans who volunteered for service with the Confederacy were given (it was actually grown in place by nanites) a cranial implant that performed various functions.  The implant allowed the AI to monitor the Volunteer's location as well as his physical and emotional status.  In addition, it hooked into the optic and aural nerves, giving the AIs the ability to send sights and sounds to the Volunteer as well as copying whatever the Volunteering was seeing and hearing.  It also could monitor impulses to the vocal cords to detect inaudible speech.  Together, this gave each Volunteer a private and secure channel to communicate with the AIs.
   When 'non-Volunteer' personnel were added to the mix, it was thought that allowing untrusted humans private access to the AIs was unwise, so each Concubine was given a collar that performed all of the same functions -except for the visual output- but anyone nearby could hear both sides of any conversations between a Concubine and an AI.  When underage (before their 14th birthday) humans joined in, they were given bracelets, although for the youngest the only functions that mattered were location and life-support monitoring.
   In early Confederacy society, the easiest way to determine the status of someone unknown was to see if they were wearing a necklace implying that they were a Concubine, a bracelet implying that they were a Dependent, or nothing at all implying the cranial implant of a Volunteer.  The external devices started as simple bands, but as time went on they became more and more ornate.

AI Operations

• Monitoring of personnel
• Responding to informational queries from sponsors and permitted non-sponsor personnel, depending on the nature of the inquiry
• Operating Confederacy technology systems, such as transporters, replicators, med-tubes, nanites and sleep trainers
• Communicating and coordinating with other AI systems in an undocumented way

AI prohibitions (Policy)

While the AIs could do all sorts of things, by policy there are a number of things they will not do. These include:

• An AI will not permit unqualified personnel to leave Earth orbit for transit to a colony (exception, the original non-Confederacy worlds established by the UN, China or others, but only in those initial ships donated to those entities)
• An AI will not allow duplication or access to material protected by copyright or identified as classified
• An AI will not reverse menopause or otherwise restore sexual functions in someone who has lost them
• An AI will not trigger a weapon to fire
• An AI will disallow direct access to other Confederacy races. Thinker weighs in on this as follows:

The AIs, treading VERY lightly and only involving races that could even tolerate discussion of the subject, got the Darjee to hold the bag as the only race required to make direct contact with humanity. Other races are not officially involved. And we all know that the AIs are REALLY running things, and have no immediate plans to expose the Confederacy as a whole to our warlike asses. Other races aren’t providing technology and materiel because, in most cases, THEY DON’T EVEN KNOW WE EXIST! The AIs won’t even tell us directly which systems are occupied, to keep us from accidentally making contact with other races. What they DO tell us is which systems are off-limits – and they enforce it by limiting navigation. We just can’t go to those places in a Confederacy vessel. The Tu'ull have broken through there, but it was their idea. The AIs have to work around it.

• An AI will not provide access to certain technologies deemed too dangerous for humans to have access to
• An AI will not allow humans to undertake actions which are likely to cause harm to Confederacy interests
• An AI will not directly cause harm to humans or other Confederacy races, with the following exceptions:
  • activating the self-destruct on a vessel about to be captured by the Sa'arm
  • altering the perception of a human to cause him to deliver harm on another human when that human has broken the law (Canon murky here, may be heretical but it has been used)

(Someday this will be a navigation template.  It will provide a bar across the bottom of each article with useful navigation links.  Until then, this is just a placeholder to get rid of all the red "broken link" indicators. -ZM User (talk) 10:00, 3 May 2024 (PDT))