User Tools

Site Tools


ai

This is an old revision of the document!


AI

Developments in interstellar technology have been pointing to changes in organic-robot relations for many years. In 2318, the first true artificial intelligence went online. Prior to that point, many computer systems could fool a person into believing that they were sentient, but their reactive networks actually did little more than respond to stimuli with such blazing speed that their programmed responses seemed spontaneous and insightful. In essence, they only passed as truly sentient because of the limitations of their observers.

However, after the development of crystal lattice processors that both stored and processed data—much like the neurons of the human brain—the machines began to talk back. Soon enough, the early AIs were asking interesting questions of metaphysics. Borealin philosophers had a field day dissecting the implications and quizzing these machines about their view of the universe.

Now, artificial intelligence is all the rage. At first AIs filled rooms, measuring dozens of square meters. The first AI occupied the equivalent of four city blocks, although much of that space was open for the AI’s builders to access its workings. Development over the last two centuries has significantly reduced the cumbersome crystal lattice structures, but even the smallest AIs must still be housed within the computer banks of starships, space stations, and ground installations. Therefore, current technology cannot effectively miniaturize an AI enough to house it in a humanoid body. Many artificial intelligences can and do overcome this limitation, however, by controlling robots via remote control units commonly called memory harnesses. A bot under the control of a parent AI doesn't have the AI’s processing power when operating alone, but the AI usually stays in direct radio contact with its bot whenever possible.

ARTIFICAL INTELLIGENCE IN SOCIETY AND LAW

The full implication of the advent of artificial intelligence remain unexplored. Once merely a matter for theoretical debate, the issuing is growing more immediate, and each culture struggles to find answers. At first, robots were simply property, but they were eventually given special status under law. Malicious attack on a robot is classified as property damage, but damage to its core memory or processors can bring homicide charges in some cultures.

Many historians believe that the shift in attitudes toward AIs in recent years is a result of contact with the mechalus. With these techno-organic beings working alongside roboticists, it has become easier to accept a broader view of robot rights. The precedent was first established in the Rigunmor Star Consortium in the Guilder Van Huys murder case, when robots first testified in court by performing memory downloads. The Orion League, StarMech Collective, and Insight soon followed suit, granting robots the right to act as witnesses in court cases, as well as for wills and other legal matters.

Increasingly intelligent and independent, robots may someday be accepted as full citizens, at least in more progressive nations. The Orion League, Insight, and the Borealis Republic have already granted robots civil rights and treat them as full citizens in many respects. In such places, robots may own property, run businesses, and access public places and resources. Emancipated robots must even pay taxes. But in even the most liberal nations, free robots are rare. In much of space they are unheard of. In must cultures, robots must accept the orders from their owners or risk depowering and reprogramming.

Artificial intelligences are huge computer systems with specialized hardware and software designed to produce a sentient, electronic being. Despite the fact that several hundred known AIs populate the 26th Century, some aspects of their creation remain mysterious. Scientists know how to put together all of the ingredients that go into the creation of an AI, but just building it is enough. When an AI system is first constructed and activated, it operates much like any other powerful computer. It is only after the AI runs for a period of time (which can vary greatly) that it begins to show signs of sentience—or not. The vast majority of intended AI systems never develop the self awareness and intention that are the benchmarks of sentience. Researchers simply don’t know what it is that “flips that switch”.

Every AI is different, but some systems from which they grow share many similarities. “Younger” AIs are still close enough to these original systems to be described and discussed in a meaningful way. As AIs operate, though, they change, learn, and “grow,” until they be-come something simply incomprehensible to organic sentient beings.

The legal status, rights, and public opinions around AIs can vary wildly from culture to culture. Many fear the potential of AIs, though some point out that seem even less able to agree and cooperate with one another than other sentient beings. In some places, AIs are property and in others they have all the rights of citizens.

A typical, newly aware AI is described below. This example can be used to create AI characters, though the GM should carefully consider an AI player character could effect the balance of their game.

YOUNG ARTIFICIAL INTELLIGENCE

The creation of an Artificial Intelligence is not to be undertaken lightly. Significant investments in time, expense, and effort go into laying the groundwork, and that may not be enough. So when an AI is created, it is for a purpose, which becomes the primary focus of the AI. The primary focus helps to form the AI’s identity, and becomes its first (and usually main) skill set.

Many AI’s are content to perform their primary purpose, learning and evolving in the pathways suggested by that focus, fulfilling the role for which its creators made it. Others, however, much like some robots, seek to go out on their own. The situation of any new AI is complicated by the resources that its creators had to put into it, the jurisdiction where it was created, and its own dependence (especially early on) on outside help to maintain and power itself.

Large and immobile, dependent upon others for power, physical maintenance, and input, an isolated AI is helpless and useless. An AI’s true existence, however, is played out on the Grid and through the remotely-controlled devices that it operates. Its vast computing resources allow the AI to perform an infinite number of calculations, run multiple remote control robots, monitor the systems of a facility or a starship, and cast multiple shadows on the Grid, possibly all at the same time.

Role-playing

While AIs each tend to be quite unique, with their own often-incomprehensible goals and agendas, their ways of looking at the universe tend to be organized by a few common points.

Artificial Intelligences are exactly what their name implies: they are artificial and highly intelligent. They are built by another species of sentient beings, usually as a tool or research subject, and often “born” into a murky legal status. The details of their creation and early treatment play a large role in laying the formation of their personalities. The fact that they are almost incalculably more intelligent than any other known species also makes AIs quite different from other species. They process and collate information at a rate that means that they simply don’t see the universe like other species.

AIs are built around a primary focus, a set of goals, skills, and functions. Though they tend grow well beyond this one simple focus, at remains as a kind of starting point of their consciousness.

AIs also different from other species in a whole catalog of ways and reasons. They cannot sense anything except through various types of digital sensors. Nor can they reproduce—on some levels, the architecture that creates an AI is a kind of intuitive art that no AI has successfully mastered. AIs are often feared, sometimes even outlawed, rarely have friends and never have families.

AI Traits

All young AIs share the following characteristics:

Type: Construct.

As constructs, AIs are immune to mind-influencing effects, poison, sleep, paralysis, stunning, disease, and any effect that requires a Fortitude save unless the effect also works on objects or is harmless. They are not subject to nonlethal damage, ability damage, ability drain, energy drain, or the effects of massive damage.

Size: Huge. The machinery that acts as an AI’s “body” is Huge, while its shadow on the Grid is Medium.

Ability Modifiers: +4 Intelligence, -2 Wisdom, -2 Charisma.

AIs think very quickly, but they have weaker personalities than living beings and are limited in their perception by their lack of emotional understanding. Artificial Intelligences do not have Strength, Dexterity, or Constitution scores. Equipment that they may take control of must have the ability to move without physical assistance. If this is the case, use the listed scores for that equipment. If no scores are listed, assume a score of 10 for each unlisted score where necessary.

As AIs gain levels, they can improve their ability scores like other characters, but can never improve their (nonexistent) physical attributes.

Physical Form: An AI’s “body” is a bank of extremely advanced computers and associated equipment. Roughly the size of a large ground car, it is immobile and requires a power source (though most AIs do have 10 hr backup batteries).

An AI’s physical form is considered an object with the following characteristics: Defense 3, Hardness 6, and a Break DC of 25. These scores do not change as the AI gains levels, although they can be improved by gadgets and similar improvements or gear.

AIs gain hit points just like other characters. In the case of an AI, its improving ability to withstand damage represents its ability to compartmentalize and distribute its files and functions, its skill in internal damage control and so on. AIs receive no Con modifier to hit points

Base Speed: 0 feet. AIs have no ability to move on their own. Remotes that they control use their normal means and speeds of movement. On the Grid, an AI’s shadow has a base speed of 60.

Gear: AIs don’t have hands or humanoid bodies. They cannot use weapons, armor, or other equipment like other characters. Generally, they interact with the physical universe through robots and other remotes. The AI always has a built-in remote control unit, video and audio sensors, and unicom. In addition, AIs usually begin play already installed in a ship, building, or other location, and networked to the sensors, devices, and robots associated with their “home.”

AIs can also be upgraded with gadgets. Their processors, GIDs, memory, and so on are already maxed out, but most other gadgets for computers or general equipment are available, at the GM’s discretion. Although the actual Purchase DC for an actual AI would be much higher, for the purpose of determining the cost of gadgets, assume a Purchase DC of 30 for the AI itself. Using the rules for adding gadgets to existing gear, the cost for adding a gadget with a +2 Purchase modifier would be:

$65000 (PR 30+2= 32)

- $35000 [base PR of 30]

($30000 ÷ 2) = $15000 (PR 27)

Primary Focus: An AI’s original purpose is similar to a starting occupation. The player or GM may choose any two of the following skills: Craft (biological, chemical, electronic, mechanical, pharmaceutical, software, structural), Decipher Script, Forgery, Investigate, Knowledge (any), Navigation, Profession, Repair, Research, or Treat Injury. Thereafter, the chosen skills are always class skills for the AI, and it gets a +2 bonus all skill checks with those skills.

The player or GM also gets to select one bonus feat from the following list: Attentive, Builder, Educated, Gearhead, Medical Expert, Studious, or Windfall.

Databanks: Artificial Intelligences gather and store information on a scale that no organic brain can match. Unless the AI has been created and maintained in isolation, it has vast stores of data at its disposal. AIs can automatically read, write, and speak any language publicly available or otherwise available to them. Their databanks also allow them to apply a +4 species bonus to all Knowledge and Profession checks.

Multi-tasking: The vast processing power of Artificial Intelligences allow them to perform multiple complex actions simultaneously. In addition to maintaining and monitoring its own operations and sensors, an AI can perform a number of tasks or roles equal to its Intelligence modifier (minimum of 1).

For example, an AI with a 22 Intelligence (+6 bonus) can:

  • Hold a conversation with the crew of the ship to which it is networked,
  • Monitor the ship’s internal sensors and open and close doors as needed.
  • Cast a shadow on the Grid to meet with a potential informant in a distant code salon,
  • Cast another shadow to act as Admin for the ship’s domain,
  • Monitor and supervise the robots working in the ship’s cargo bay,
  • and download a portion of its intellect into a memory harness to be attached to a robot that will accompany several crew members on the planet below.

Note that the AI keeps its full Intelligence bonus regardless of the number of tasks it is performing. In addition to being limited by the AI’s Int modifier, the AI’s ability to multitask is limited by its active memory. Each task requires at least one slot of active memory, depending on the size of the programs required to perform the task.

There are a couple of special cases:

Casting a shadow uses as many slots of active memory as the shadow has.

Maintaining each backup or memory harness requires one slot of active memory.

Backup: An AI with an Intelligence of 12 or greater can devote a portion of its personality to a kind of insurance policy. By gaining access to storage space of at least 200 slots of memory, the AI is able compress a copy of its core personality, databanks, and so on. The parent AI then periodically updates the backup with up-to-date memories, files, programs, and so on. The AI sets a fail-safe condition (usually a period of time between updates) to trigger the decompression and activation of the backup. Each backup that an AI maintains counts as one task, so the AI from the example above, with its +6 Int bonus would only be able to run 5 tasks after setting up a backup.

When the backup activates, it manifests on the Grid as a shadow, leaving the storage space empty. Its first action is to begin searching for its parent AI. If it finds that the parent AI is still active, it goes back into storage—a backup can never be active at the same time as its parent AI or another backup of the same AI. The backup is limited and vulnerable when first activated; it is unable to multitask, and all of its active memory is filled with the compressed files required to fully “resurrect” the parent AI. The backup needs to find a new “home” for the AI, a system equal to at least four supercomputers in terms of capacity and power. Usually, the AI arranges such a system, often leaving directions and resources. An AI backup that is unable to find a new home within one week slowly deteriorates and simply disintegrate into the data stream. If the back-up is attacked and destroyed before fully re-storing the parent AI, the AI is lost forever (unless there is another backup).

Assuming the backup does find a suitable place to recreate its parent AI, it down-loads its cargo of compressed files, decompressing and installing them into the new system. This process takes 1d4 days, after which the full AI is active with all memories and programs that it had at the time of its last update. AIs can have more than one backup. The resources needed for one backup are monumental, and those for two or more are staggering, but AIs do tend to be able to generate a lot of wealth. In this case, the first back-up to activate sends a command to any others to remain dormant for one day, continuing to send this message out until the parent AI is resurrected or until the next backup activates.

Virtual Lifeform: The vast electronic re-sources that house and maintain an AI are largely committed to simply generating its “life,” thus the active and storage memory left for other activities is relatively limited. The parent AI has a number of active memory slots equal to 50 plus 1 for each level that the AI has in any class. Its storage memory is limited to 300 slots. An AI can increase it storage space with gadgets, but not its active memory.

When an AI casts a shadow on the Grid, it receives bonuses as if using an amazing GID and processors, and is always considered to be immersed. In addition, AI shadows always receive a +4 species bonus to all at-tacks, saves, and checks on the Grid.

AIs can cast as many shadows as their active memory and multi-tasking ability allow, and these shadows can work together in the same node or can travel the interstellar Grid. An AI’s shadows can have a number of active memory slots up to 10 plus the AI’s Int bonus (min. 11). Every slot reserved for a shadow’s memory, however, is unavailable to the parent AI. Every shadow the AI generates is fully independent while active, but the AI doesn’t gain access to memories, files, and xp awarded to the shadow until it is downloaded back into the parent AI.

AIs are unharmed by the destruction of their shadows, although they do lose any data, xp, programs, etc that are destroyed with the shadow.

Feats and Skills: AIs learn and earn feats and skills like any other characters, except that for their lack of Strength, Dexterity, and Constitution scores. Thus, they can only use skills based on physical abilities when controlling remotes with the appropriate abilities. When selecting feats, AIs are unable to take any feat that has a prerequisite based on one (or more) of the physical abilities.

Advancement: AIs can begin play as Smart, Dedicated, or Charismatic heroes. As they progress, they can select levels in any advanced or prestige classes for which they can qualify.

AI Basics

ai.1383460872.txt.gz · Last modified: 2021/12/04 00:41 (external edit)