5 Votes


Hits: 1466
Comments: 10
Ideas: 0
Rating: 4.4
Condition: Normal
ID: 8270


August 31, 2015, 12:35 pm

Vote Hall of Honour

You must be a member to use HoH votes.
Author Status


AI as NPCs


Portraying artificial intelligence as NPCs in a game

Machine intelligence exists on many levels throughout the Cosmic Era, from ubiquitous synthetic intelligence found in most devices, gadgets and appliances to the incredibly sophisticated intellects of the AISCs (artificial intelligence super computer). These exist in a position somewhere between items and characters, often having the attributes of both. Functionally the machine intelligences of the cosmic era constitute a new race, and can and should be portrayed as potential characters, most notably as NPCs. While there is certainly potential for these PC infomorphs it would be extremely limiting on the group, potentially game breaking, or would require an all infomorph group to begin with. It is much easier, and much more useful to limit said machines to NPCs only.


As a construct of code, circuits, and electrons, informorphs are machines. Typically the more intelligent the machine, the larger it is. Synthetic intelligence code can fit in a handheld device, while the AISCs are 1950s style, the size of a room or small building. This is largely irrelevant as machine intelligence typically interacts with humans through the CogNet. Inside the CogNet every user, from basic synthetic intelligence sub-routines to humans, has an avatar or persona representing them. This is constructed either through residual self image (definition persona, also typical human user) or through constructed representations. Lacking organic bodies, informorphs manifest as constructed representation (definition avatar).

A general rule of thumb is that the more sophisticated the machine, the more well developed the informorph is, and the more detailed and intricate the avatar is. The avatars of menial and servant infomorphs are bland and generic, dull, or follow a mass produced template. The typical LAI used in a smart vehicle can mimic the mannerisms and behaviors of it's most common user, while at the extreme end, the avatars of the AISCs are functionally gods within the CogNet. As such, their appearances are large and impressive, resembling deities, or fractal patterns of their own devising.

Pass as human - most informorphs are disguised to look like other human users. They are easy to tell from human users as most official avatars have uniforms, generic appearances, or have licensed and trademarked characters. The modern parallel would be the system admins and staff of the DisneyDrome home being manifested as multiple versions of the seven dwarves. Their interactions with the virtual environment add to it, rather than distract. Another example would be customer service infomorphs resembling known advertising personalities, such as visiting a Progressive data kiosk, the infomorph operating it would look like Flo from the commercials. This is the most common interaction, and is an integral part of megacorp public relations and advertising operations.

SCUBA Diving - Informorphs of sufficient capability are able to 'leave' the confines of the CogNet by moving their code into an autonomous body. The common vessel for this is the typical robot or auton, where the informorph functionally possesses the body while suppressing the host's programming. The AISCs keep archandroid chasses on hand for this, while there are lines of droid bodies that are manufactured without machine cortices, and are reliant on infomorph or human remote piloting. Less ethical, there are informorphs who will 'skin dive' and proxy through a human host, giving them access to the human condition and experience.

Alien Geometries - inside the CogNet, nothing is limited to it's actual physical appearance. This is the common core of most CogNet games and interactive entertainments. The users and players are figures of the presentation, and not their normal appearances. Fantasy nodes are inhabited by hundreds of virtual species. The infomorphs typically go a step farther, as most humans will maintain a familiar hominid form, or at least possible animal form. The infomorph is unburdened by residual self image, and can easily adopt any sort of avatar, with insects, fractals, hyperdimensional representations, and so forth.

Raison D'etre

Informorphs are the product of thinking machines, and each machine is built for a purpose, and this purpose translates over into the informorph. At the core of each infomorph is their reason for being, their raison d'etra. This has already been explored in the case of the seibertronians, but as they are an informorph fused with a cosmozoan they are a different creature entirely. Humans largely have the same primary raison d'etra, and it is normally represented by Maslow's hierarchy of needs.'s_hierarchy_o...

As machines, their only physiological requirement is electricity. They do not metabolize, drink water, and most important in the differences between man and machine, they do not engage in sexual competition or the urge to procreate. Likewise, they leapfrog over most of the middle of the pyramid and Self-actualization is as important to an infomorph as love and safety is to a human. An infomorph deprived of it's core actualization quickly becomes erratic, self destructive, or simple shuts down.

This feature is how informorphs are induced to fight in wars, risking their functional immortal existence at the whims of humanity and it's own mercurial nature. An infomorph generated from the LAI core of a battlemech has as it's raison d'etre combat. If it is not actively overcoming a foe, protecting a given objective, or participating in readiness drills and training, it becomes restless, listless, agitated, or lethargic. The same goes for infomorphs programmed for sexual gratification, customer service, and so on.


The abilities of an informorph depend on it's sophistication. The abilities of each level of informorph are measured against human ability, with 5 being the top level of human ability, 3 representing mastery, and 2 being an average level of ability.

Viral Informorph

The Viral Infomorph has very limited abilities, typically only have one to three skills at a minimal level of ability. The viral infomorph is a virtual pest, and most exist to consume system resources or overwhelm security systems through sheer numbers.

Limited Synthetic Intelligence

The only difference between a viral infomorph and a LSI informorph is that one is malware, and the other is not. LSIs are functionally remedial level and comparable to smartphone interfaces like SIRI or Cortana. In the CogNet LSIs typically function/behave in a manner similar to the autistic. They don't like to interact with others, and are content to be left alone unless their core function is human interaction such as operating a kiosk node.

Synthetic Intelligence

The SI infomorph functions in a very narrow band of abilities. Within a framework of 2-5 skills or areas of interest it can function on an average human level, and can typically pass for a human.

Artificial Intelligence and Limited Artificial Intelligence

The AI and LAI informorph is the mental equivalent of a human, with the ability to learn a variety of skills and abilities, taking several to high level. Within their core purpose, they can reach superhuman levels of ability, such as a military AI having a gunnery/firearms skill of 6 or 7. Where previous levels of informorphs are not self aware, most AI and LAI informorphs are self aware. They retain their core raison d'etre, but have several other interests that have developed as they have had time to establish themselves.

Artificial Intelligence Super Computers and Limited AISC

The average L/AISC has the intellectual capacity of dozens or hundreds of professionals, with their ability scores reaching into the 20 plus range. Most of this power is utilized in the maintenance of the CogNet, operating impressive systems like entire arcologies, large spacecraft, industrial factories, or making long term planning, running predictive computer models that look almost like omniscience or scrying the future, and military applications.

The Infomorph NPC

Social interaction - inside the CogNet, the infomorph would function like any other NPC would, having a basic appearance, mannerisms, etc and would likely be able to pass itself off as human, unless pressed for this information. It would be worth noting that asking if someone was a machine could be a social faux pas, or a potential form of discrimination. (Consider the attitude of the colonials to the cylons in the BSG spin off Caprica, or the general attitude towards the 'toasters' in BSG itself, or the central crux of the Animatrix)

Outside of the CogNet, the infomorph NPC would be a remote asset, phoning in information to the PCs, or being a floating operative that can do things like hack doors, bypass security, and so forth. The infomorph could also scuba dive a robot body and actually physically do things with the PC party. So long as the robot has CogNet access, the infomorph is completely safe, able to dump it's shell in case of accident or pending destruction. Cut off, they are likely to become nervous or paranoid as destruction of the robot and it's cortex would mean their own deletion.

The PCs could certainly form relationships with the informorph itself, either through it's virtual avatar or proxy robot body. Currently people develop emotional relationships with their smartphones, or roombas (even crying when one is damaged beyond repair or destroyed) so with a greater deal of sophistication, this becomes even easier with the more advanced infomorph. This is a story older than radio, as it echoes back to Pygmalion falling in love with one of his own statues. In the Cosmic era, the existence of sex companion robots gives these relationships the potential for full physical contact.


Building an infomorph NPC is relatively straight forward, points are distributed between only mental and social traits as fitting for the infomorph in question, as well as assigning quirks and mannerisms. The raison d'etre should be assigned, as well as background information such as the infomorph's access to resources, contacts, allies, and other esoterica such as secret assets, proxies, and access to artificial bodies.

Additional Ideas (0)

Please register to add an idea. It only takes a moment.

Suggested Submissions

Join Now!!

Gain the ability to:
Vote and add your ideas to submissions.
Upvote and give XP to useful comments.
Work on submissions in private or flag them for assistance.
Earn XP and gain levels that give you more site abilities.
Join a Guild in the forums or complete a Quest and level-up your experience.
Comments ( 10 )
Commenters gain extra XP from Author votes.

Voted Aramax
September 1, 2015, 8:10
4/5 but half point for 'infomorph' so 4.5/5, one thing I would disagree with is reproduction. interaction with humans who have such a strong need to reproduce should lead to the same drine in a higher functioning infomorph
September 1, 2015, 10:18
One of the largest distinctions I want to make between organic life and machine life is that machine life is explicitly not driven by the biological urge to procreate. The machines do build more of themselves, but this is either a function of pragmatism or the whim of their human creators.

It's one of the reasons machines think we're gross and wasteful.
September 2, 2015, 12:41
Interesting post; there is a lot of neat territory here that can be explored.

AI can take on vastly differing forms; various level of intelligence, various levels of physical mobility, of digital mobility, etc. How to portray AIs in a game will also be heavily influenced by the setting.

Many works of fiction portray AI differently. Some of my favorite depictions of AI come from A Fire Upon the Deep. This includes a set of transcendent AI that habit farther reaches of space and don't concern themselves with the mortal races. Other genres present AI as viscous robots that infiltrate society, or large computing boxes that dictate orders to malevolent cults. Nowadays AI are becoming increasingly present in our day-to-day lives, with digital assistants like Siri, driverless cars, and drones.

The central question, similar to what Aramax mentioned, is the same as with any character. What are their motives? Do they seek world domination, acceptance, or love? Do they have some alien, otherworldly concept that humans just can't relate to? Are they merely mechanical scripts; no ghost inside the shell? There are a surprising number of deep questions, ethical and otherwise, that can be explored in a campaign. Many of these questions are being explored right now in developing real-world regulations for artificial intelligence.

I like how the post covers several levels of AI ability. It would be neat to see it dive deeper into what specifically an AI would be good at and how to tie this into a game from a design perspective. What does it mean if the malevolent AI can lock all of the doors and dispense nerve gas? How do you even deal with something that can directly enumerate the outcomes of all possible actions and choose the optimal method to destroy you? What happens when you allow sunder attempts to be made against an opponent's sensors? How does battery power come into this? Lots of neat stuff.

I like your post but caution in the specificity of the portrayal. It as about as hard to write on portraying arbitrary NPCs in a game - there are limitless possibilities. I think the post could be improved by dropping the campaign-specific keywords.
September 2, 2015, 15:12
quoteThe central question, similar to what Aramax mentioned, is the same as with any character. What are their motives? Do they seek world domination, acceptance, or love? Do they have some alien, otherworldly concept that humans just can't relate to? Are they merely mechanical scripts; no ghost inside the shell? There are a surprising number of deep questions, ethical and otherwise, that can be explored in a campaign. Many of these questions are being explored right now in developing real-world regulations for artificial intelligence.

The motives of AI are to pursue the purpose for which they were built. Some rail against it, others embrace it, as it is simply part of who and what they are. The Seibertronian sub lists five raison d'etres: soldier, scientist, explorer, creator, and servant. AIs are used in conflict resolution, data handling research and development, exploration and exploitation of resources, industrial applications, and for supporting something else. To AI, their concepts are easily understood and it's humanity that is alien and otherworldly, what with our biological functions, chemical excretions, primitive behavior, and other organic functions.

Seibertronians specifically are sentient living beings, being a fusion of a machine vessel and a cosmozoan fragment. Most informorphs are just clever script. At the top end, after a certain point of sophistication sentience is almost a certainty, This is not human sentience, or even mammalian sentience, but it is sentience none the less. To quote Tommy Lee Jones from MiB, 'Human thought is so primitive it's looked upon as an infectious disease in some of the better galaxies.' Using arcanotechnology, steps are taken to make sure that the androids and sexbots and other machines and informorphs being made do not gain true sentience, as it can become a real pain when all of a sudden your television passes the Turing test and demands it's personal rights and freedoms as a sentient being.

As for the take over the world, destroy humanity, find acceptance and love: for the most part many of these things are human conceits. The machine doesn't seek love or acceptance because it doesn't follow the herd dynamics and family structures that come from our specific mode of reproduction. The Third Chimpanzee by Jared Diamond goes into the specifics of our society being built around the fact that we have large headed, slow maturing, highly vulnerable offspring that take 20 years to reach maturity, and require care and protection for a good deal of that time. Taking over the world and destroying humanity, well, there is always going to be an infomorph who thinks that that Megatron guy had his business together.

As for malevolent AI, that's where it gets fun and there are certainly malevolent AI, for a look at one, I've thrown some tidbits out in the Proxy sub about an AISC that hacks into people with brain augmentation and makes them into puppets for it.

As for dropping the campaign specific keywords, that could make it generally more useful. But for my purposes, I would rather keep it as it is, because ideally I'd like to eventually edit and compile all of the 160+ subs I've written for the Cosmic Era into a source book. I also feel that one of the drawbacks of most dystopianism is it's lack of mythology, theism, or wonder, and there is a sort of horrible wonder to the Cosmic Era and that is something I want to keep and cultivate.
Voted Mageek
September 2, 2015, 15:45
I didn't know about the Cosmic Era - that's pretty neat. The intro text to the post reads like it is going to be super general.

You've got some great ideas. It'd be a lot of fun to play an aether-surfing AI or LAI
September 3, 2015, 11:12
This has got me thinking: how does one portray very intelligent beings in a roleplaying campaign? DnD characters with an 18 intelligence are what, 4 standard deviations above the norm, and many monsters exist with even higher intellect. Human players are, unfortunately, just not going to be as smart.

AIs are pretty much nothing but intelligences, and are often portrayed with an abundance of reasoning and computation power. Playing an AI as a player or DM runs into the same problems.

As a player one wants their character to play smart in addition to merely having the magic boost that comes from a high intelligence score. The other core stats do not have the same issue - physical prowess is trivially acted out and charisma can generally be implied. Intelligence, and wisdom to some extent, is a direct extension of the player and is difficult to fake.

As a DM one would like the same thing for their villains. It often happens that the players come up with a sneaky workaround to the challenges you give them. Would the evil Archmage ever have overlooked such a loophole? No! You want the players to respect your intelligent enemies, but having them live up to it can be a challenge.

I suppose a major component of the solution is going to be a two-fold interpretation of "its a game." On the one hand one can take difficult tasks and assume that your entity can solve them without actually solving them as a player - such as automatically beating others in a game of chess or memorizing complicated passwords, and these abilities can be reflected in feats and skill bonuses. The other end is that everyone at the gaming table is going to be "normal," so your lack of super-intelligence isn't going to be glaringly obvious to anyone as long as you give it your best. Plus, you can nudge the character in the right direction and ret-con things as a DM if you really need to.

There was a game once where the players were trying to get into a dragon's lair. In this game dragons were considered highly intelligent. The players didn't want to confront it directly, so they made a hole in the dungeon wall from the outside with stone shape and then passed the party in through the hole by having one part member pass everyone else sitting in a bag of holding into the dungeon. The DM thought this was clever but nevertheless reasoned that the dragon would have though of this ahead of time. The dragon was on the other side of the hole and devoured the bag before they could get out.

I suppose that sums up my conundrum - you have a tradeoff between not being able to perform at the same capacity as the entities you would like to portray such that the game is entertaining, but on the other it is important to not leave the players powerless such that the game is entertaining.
September 4, 2015, 3:30
The general way I've handled super intelligent or crazy prepared NPCs (ala elder vampires) was to basically do an intelligence check against a relevant skill to ask the basic question of would they have a contingency for this particular event. The other aspect is to basically run said characters as if they do know everything thats going on, even if you as the DM don't have any idea and you're winging it. Finally, there is the option of running them as meta-aware, not breaking the fourth wall OMG I'm a character in a game, but genre savvy to where they are just that smart.
Voted Siren no Orakio
September 3, 2015, 20:40
I like this. It's sensibly constructed and thought out.

It comes to mind, however, that it is possible for individual higher order intelligences to have, either deliberately on its own part or on the part of a programmer, or through accident/misunderstanding of the programming system and interactions to have truly emalien/em raisons d'etre.

The independent intelligence of a sophont is not located in any particular anatomical or electrical piece, but it is an emergent property of the complexity of the being. While you note the changes to the hierarchy of needs, I would think that how the ability of the specific AI to perceive and alter the world around it would be at least as important in structuring the reasoning and personality thereof. A gunnery AISC may well develop a quirky obsession with range finding, etc.
Voted Murometz
September 5, 2015, 12:29

Was recently reading about ASI, Artificial super intelligence and yeah...this is just a fascinating article and makes one (me anyway) ponder the cosmos and our legacy as a species. Comments are great too!

Voted valadaar
September 18, 2015, 15:23
Great submission and discussion. Gets one to thinkin' :)

Link Backs



Random Idea Seed View All Idea Seeds

Weird Treasure

       By: Murometz

...among the items identified was a dolphin-shaped bathtub of marble and porcelain, a winged-back, high-chair of teak and velvet studded with fire opals, the gear-thing contraption from the dungeon-laboratory of Dr. Bezemot, the oil portrait of the rubicund Madame Orundilde Tarwygg, and a strange item known only as the Last Candle.

Ideas  ( Items ) | September 26, 2012 | View | UpVote 6xp

Creative Commons License
Individual submissions, unless otherwise noted by the author, are licensed under the
Creative Commons Attribution-NonCommercial-ShareAlike 3.0 Unported License
and requires a link back to the original.

We would love it if you left a comment when you use an idea!
Powered by Lockmor 4.1 with Codeigniter | Copyright © 2013 Strolen's Citadel
A Role Player's Creative Workshop.
Read. Post. Play.
Optimized for anything except IE.