Artificial Intelligence (A.I.)

Background
Epoch:Human is based upon the timeline of our current reality, projected through to 2150. So events and the state of the real world today potentially still have some echoes in the history of the time period of Epoch:Human.

This article does not describe the fundamental technologies that computers and networks are built on in Epoch:Human. It is assumed for the purposes of describing the existence of A.I. that the necessary hardware and network/communication tech exists which can be used to build computing architectures that can handle the requirements of A.I.

A.I. general information

 * AI in the world should not be the full-on self aware SkyNet or Minds (Iain Banks "Culture" novels) seen in much of the distant-future sci-fi genre.

High-A.I.

 * High-A.I. should be restricted to high-level powerful entities such as governments, military agencies, large/mega corporations and the like.


 * High-A.I. while not actually being truly self-aware, is of a very sophisticated nature. Advanced hardware/software constructs that have some level of limited self-evolution ability.


 * High-A.I. constructs can *simulate* intelligence to a high level, but are not perfect at it. There are some flaws that give away their nature: poor emotion handling; likely to be too literal in interpreting requests and information; and other similar problems.


 * High-A.I. can easily simulate a high-level human intelligence, and in many ways can exceed it. Analytical and data mining abilities are far beyond that of a human.


 * High-A.I. is not fully trusted by its operators, and is carefully monitored and controlled. Potentially lethal emergent behaviour is not to be tolerated.


 * Government agencies of some type may play a part in the regulation of High-A.I. systems.

Low-A.I.

 * Low-A.I. (normal A.I) is more primitive in nature compared to High-A.I.


 * Low-A.I. systems of varying levels of power/ability are widely available and occur in everyday use by most citizens.


 * Low-A.I. systems often vary in their intelligence simulation from the equivalent of a small rodent upto an "average" human.


 * Low-A.I. systems are limited by the resources available to them such as the hardware they are running on. Depending on the hardware capabilities, a Low-A.I. could mesh with other local hardware resources to increase available processing power, and thus be able to be more effective in complex tasks. Of course, meshing with unknown/untrusted devices brings the risk of running part of the AI simulation on a compromised system, so may lead to unexpected problems.


 * Low-A.I. cannot be made into High-A.I. simply by doing the equivalent of throwing more CPU/RAM resources into it. The whole architecture of a Low-A.I. system is simply less sophisticated and less capable than a High-A.I. system.


 * Low-A.I. systems are often built to perform a specific job/task.

F.A.Qs

 * Why are we not having fully sentient self-aware True-AI? - [1] the world is set in 2150, not 2450, and is an extension of our real timeline, not an alt-history. [2] the tech behind A.I. should have some justification for existing as a logical extension of current-day tech. [3] It's not as full of potential: 2150, AI still has some rough edges, efforts maybe underway to perfect the tech; but the potential for things to go wrong may be fairly high ...


 * Ok, so how realistic is the High-A.I. intelligence simulation? - I would say it can be quite good, but often with some noticeable flaws. Consider this (as an extreme): What if Google + Facebook + Amazon + Wolfram Alpha had a love-child with Dr Sheldon Cooper (from The Big Bang Theory)


 * So what is the intelligence simulation of Low-A.I. like then? - Think of something like Apple's Siri powering your toaster, with a bit more intelligence, enough to get the job done, but still lacking in emotional context handling and not great at dealing with illogical concepts. Low-A.I. varies in ability as described above, depending on the job it was designed for. The higher end of Low-A.I. will still have it's inherent flaws, but it will be improved compared to less advanced Low-A.I.


 * How does this mesh thing work then? - Mesh networks are interesting things and will be included in the Tech Lore. In the specific case of wearable/implanted tech, I can see such devices meshing to create a BAN (body area network). As a result all of the meshed devices would benefit from additional processing resources.