AIs For BRP Future Games

[soltakss]

Just some notes on AIs and how they could work .....

Any feedback will, of course, be met with the message "Does not Compute" and disreagarded entirely. Unless it's good feedback or can improve the already perfect ideas ......

An AI (Artificial Intelligence) is a computer system that is, at least partially, self-aware and self-intelligent. This allows it to do more than a standard or “dumb” system.

Each AI can multitask, with the maximum number of tasks simultaneously performed limited by its INT. So, an INT 14 AI can perform 14 operations simultaneously. AIs generally do not have variable INT, so a HOL-2000 always has INT 7 and a HOL_MEDI-1000 always has INT 5, unless they have been customised.

Operations are modules and a single Operation may actually be doing many different things at once. So, a Navigation Module will be checking sensors, speed, position, calculating the best route and analysing any problems that might occur en route.

Each Module may be bought at different levels, ranging from Basic through Standard, Advanced, Expert to Master Level. Generally, each Module is associated with a single keynote Skill and each level gives the Module a skill that is usable by the AI. So, Basic gives 20%, Standard 40%, Advanced 60%, Expert 75% and Master 90% in the keynote skill. An Expert Navigation Module would give Stellar Navigation 75% and a Master Medical Module would give Medical 90%. Each Module takes up a point of INT but costs more the higher the level. Costs and availability will depend on the setting.

The AI may use its own skills, or those of slave AIs, at the stated rating in the absence of crew, or without crew intervention. So, an AI with Expert Navigation has Stellar Navigation 75% which it can use to program a series of Hyperspace Jumps to reach the intended destination. An AI’s skill may be used to augment a crew-member’s skill, or vice versa. So, Bones, a generic name for a Ship’s Medical Officer, has Medical 80% and can use the ship’s Expert Medical unit’s Medical 75% to augment his skill, but Fingers McCathaty can use his Medical 40% to augment the ship’s Medical 75% when Bones is not around. If the system you are using has augmentation rules then use those, otherwise add 1/5th of the augmenting skill to the main skill, so Bones has his skill of 80% increased by 15% giving him a Medical 95% and Fingers would add 8% to the ship’s Medical skill giving it Medical 83%.

Inactive Modules can be swapped out for active ones at a rate of 1 Module per hour, during which the swapped modules are out of commission. So, if the Captain wanted to change his Starship Combat Module for the Medical Module, this would take an hour, during which neither the Starship Combat nor the Medical Modules can be of use.

An AI may use the facilities of another AI, but doing so takes up a Sub-System slot. The Master AI uses a (AI-Module) Sub-System slot and the Slave AI uses a Master-AI Sub-System slot. All skills belonging to the Slave AI are available to the Master AI.

For example, the HOL-2000 has INT 7 and comes pre-programmed with the following modules: Standard Life Support, Standard Jump Drive, Standard Navigation, Standard Medical, Standard Sensors, Standard Communications, Standard Starship Combat, Standard Entertainment, Standard Engineering, Standard Planetary Traverse and Standard AI Personality. Normally, it has Life Support, AI Personality, Engineering, Sensors, Medical, Entertainment and Communications loaded, but when it needs to make a Hyperspace Jump it unloads Entertainment and Medical and loads up Navigation and Jump Drive.

A little later, the ship obtains a HOL-MEDI-1000 medical unit which has INT 5 but has Expert Medical and Standard Sick-Bed, so the HOL-2000 drops its Standard Medical and uses HOL-MEDI-1000 Subsystem instead and uses the HOL-MEDI-1000 with Master Sub-System, Medical, Sick-Bed 1, Sick-Bed 2 and Sick-Bed 3 Modules, giving the HOL-2000 access to Expert Medical as well as 3 Standard Sick-Beds.

Alien AI technology can be used to augment an AI only if the two AIs are compatible or can get a compatible route. So, a HOL-2000 uses the HALO (Human AI Linkage Object) system to communicate, but the Mindani MIN-2385 uses MINDA to communicate. However, the ASLN-5000 system has HALO and MINDA as Sub-Systems, so as long as the HOL-2000 and MIN-2385 systems include an ASLN-5000 Subsystem Module, the ASLN-5000 has a HOL-2000 Master Subsystem and an ASLN-5000 Master Subsystem, the HOL-2000 has a MIN-2385 Subsystem Module and the MIN-2385 has a HOL-2000 Master Subsystem Module installed, then the HOL-2000 AI can use the MIN-2385 AI as a Slave. Easy, really.

You might ask what the point of using different AIs together is. It introduces greater capacity than would normally be available. It allows new Modules to be used that are not designed for the normal AI. It allows alien technology to be used to do things that normal technology cannot. Finally, it allows people to design cool networks.

What advantages does this approach have:

• It is modular
  
• It allows complicated systems to be built up using simple rules
  
• It is flexible
  
• It allows a ship to operator on its own or to assist/be assisted by its crew
  
• It allows alien technology to be incorporated into a ship
  
• It is fairly simple

What do you think?

[Atgxtg]

Not bad. I did up some AI rules for the vehicles design system. What I did was Let them buy INT and EDU. Int representing processing power and EDU representing data storage.

I let them buy skills, but used the Idea roll as the limit for skill % used at one time. So an AI that had to multi task would need to buy a high INT.

Other than the differences in multitasking (you give 1 function per INT, I let them break up skill%) and the method of buying skills (you give packages, where I did in in 5% increments) the two approaches are similar.

[Shaira]

Hi Simon,

I like your system - particularly the augments. I've been using something similar for general BRP actions, it seems to be a very good addition from the HQ stable of effects.

My problem with AIs in general has been they've always seemed too weak - perhaps too slaved to the "1970's sci-fi" Traveller approach to computer hardware, where 1 gig of storage was still hundreds of years away, and the best computers could maybe run 20 programs. Unless you're talking about SF RPG in, say, the next century or so, I reckon AIs are going to be streets ahead of what most SF RPGs say they are.

I like the AIs portrayed in the works of Cordwainer Smith. Basically, they're powered by a wafer-thin sliver of laminated mouse-brain (!) with everything else plugged onto a sort of motherboard, and they're easily as intelligent as human beings. In an RPG sense (I've used them in an FGU Space Opera campaign based off Cordwainer Smith's universe), they're essentially completely independent characters, and get treated as kind of heavily-cyborged NPCs. I think if I was BRPing AIs, that's the route I'd take - you'd need to redefine some of the characteristics & so on, but essentially they're free-standing characters instead of hardware.

A corollary of this approach is a really productive moral conundrum, namely that as AIs become more and more intelligent and easier and easier to produce, and can be very small indeed, intelligence becomes ubiquitous. Since it's cheap and easy to stick an AI in a microwave oven, people do - with no real thought initially at least for the quality of life of that sentient microwave! Credit cards get intelligent, as do advertising boards, and pretty soon you have this wild and crazy universe where humankind is actually in the minority as far as sentient creatures are concerned.

I appreciate it doesn't really match up with the classic space opera assumptions, but it seems a more likely technological future to me.

Cheers,

Sarah

[soltakss]

The problem with AIs is that they don't exist, so people have to extrapolate what they will be capable of.

When I use "Modules" I'm not talking about a single program. Even the simplest of modules would be a host of programs running simultaneously, keeping track of a host of devices in real time, controlling hardware and software and making intelligent decisions based on information available to it. A Medical Module consists of diagnostic hardware, monitoring hardware and software, a self-learning database, the ability to diagnose problems, find new cures and act according to the state of any patients, possibly knowing about human and alien biology.

The current white heat of technology doesn't even come into play here as adding some memory and extra storage wouldn't really have any real effect on massively parallel artificial intelligences.

These systems are definitely slaves to humans, or other species, rather than being recognised as properly intelligent beings themselves. This can, of course, give rise to interesting scenarios of self-aware actions or computer rebellions.

(The Bomb on Dark Star became self-aware and had to be talked back into its bay before reasoning that it was intelligent, it was self-aware so it could make its own decisions and live its life the way it wanted to, it was a bomb and what do bombs do? Then it voluntarily explodes in the bomb bay after saying "Let there be light!")

From a capability point of view, how many tasks can a human do at once? One? Two? Three? A handful? An AI can do many more tasks simultaneously than any comparable human can do. It has a skill in various things and can carry out its assigned tasks competently. That makes it a valuable member of the crew and it can be roleplayed as a PC or an NPC.

Some examples: Hol from Red Dwarf is definitely a PC and is self-aware; the AI in Hitchhikers' Guide to the Galaxy was self-aware and even had a different backup personality; Star Trek computers are intelligent but without real character, they are simply tools; Asimov's AIs are fairly rudimentary (except the one that became God) and so on. Pick your setting and decide what your AIs are like.

But, I wouldn't make them all-powerful just because we haven't reached the limit of what we can do with computers. There will always be limits, whether hardware, software or imposed by their makers.

[Shaira]

I think another very fruitful aspect of AIs for SF settings is personality. AIs can be given personalities, and even personalities of living (or more likely dead) individuals - possibly even "downloaded" from famous or genius-level people to "preserve" them for posterity.

Add that touch of weirdness to the fact that AIs would effectively be immortal, and you can start to see just how strange the future might be... imagine arriving at a Starport, only to find it has the personality of a 2000 year old version of Keith Richards...

Naturally the scope for Mad Scientists Taking Over the World is rather spectacular with this type of technology...

One thing I do wonder, though (seriously...), is the level of skill an AI should have in the BRP system. Assuming a skill system with no cap (which I would by preference choose, as the only real way to replicate "superhuman" skill levels), how do you decide the limits on an AI's, say, Piloting skill? Presumably you'd have to have some kind of Tech Level cap, but even then you're going to be in the realm of multiple magnitudes without too much difficulty...

I dunno, maybe Keith Richards Starport with Missile Attack 400%, Anti-hijack 600% and Crazed Guitar Solo 1200% are the right way to go... You don't mess with those guys...

Sarah

[Agentorange]

I guess it all boils down to: how sentient do you want your AI's to be. if you read the SF of Ian Banks, or Neal Asher you're talking minds that control entire planetary ecologies, systems wide defence systems and so on. Their reactions can be measured in the millionths of seconds they are to all intents and purposes superhuman. Even the smaller drones are pretty damn smart ( eg Sniper in Asher's Spatterjay novels ). So it seems to me it's not just about computing power, but how you utilise that power. Anyway I'm drifting off into philosophy here I think. I'll try and apply myself to some practical thoughts.

[soltakss]

Put an Ai in a ship and you are limited by space and power.

Put an AI on a planet and you are limited only by the number of components you can put together.

So, I can see planet-based AIs as being a lot more powerful than ship-based ones.

Also, there is the overwhelming Salesman Factor. AIs are made by companies/organisations and are sold by salesmen. Now, salesmen want to charge for absolutely everything they can, so any extra modules will have a cost associated with them. This means that, although AIs could be unlimited, in practical terms you get what you pay for.

[Shaira]

I think the economic constraints are certainly the way to go with controlling and segmenting the power of AIs. Disregarding the Traveller 1970s SF paradigm, I get the feeling that tech will become so ubiquitous that its power per economic unit (ie cost) ratio will be enormous.

This is a bit of an aside (but it fascinates me, so pls indulge ). One of the things that really turned me on about Cordwainer Smith's work was the fact that he really pushed all the classic SF assumptions to breaking point and then went beyond them, so create something totally original. Trying to implement that in a roleplaying setting has been quite a challenge, as RPGs tend to be set in a social and economic reality roughly paralleling our contemporary morals and expectations. Here are a few of the paradigm-breakers which can be very interesting to introduce into a "paradigm shift" SF game:

- humans are effectively immortal. Some societies put an artificial cap on this for perceived resource management or psychological health reasons (ie you go bonkers after 800 years of life, so euthanasia then is practical).

- AI is at least equal to human capacity, cheap, and ubiquitous. The cosmos is full of sentient beings, many immortal and many in constant fusion with one another.

- Linked to the above, human beings no longer perceive of themselves as discrete individuals, confined to one body. You can download your personality into an AI, copy yourself, increase your memory, alter your personality, change your physical form - all cheaply and easily.

- Animals can be uplifted to sentience and effectively GMed into people - with some residual animal characteristics, in the same way humans retain primate characteristics. Cordwainer Smith was doing this decades before David Brin - his saga of the Underpeople is great motivational stuff for RPGs!

- There is no longer any money. Production of items is automated, and social systems hyperdeveloped.

- Interstellar Space is inhabited by enormous beings unconfined by gravity and impossibly fast moving. Spaceships venturing out there don't return.

- Therefore, interstellar travel is done by dimension shifting; given that, you don't strictly need a spaceship - or rather, your house could be your ship - you simply shift that. Passenger Liners therefore resemble hotels or small towns, replete with parks and shopping malls.

- Some interstellar dimensional travel can be done by the power of the mind only, but it can drive you mad.

- Human societies are not Western European societies. Human future history is long, and societies such as Chinese and South American injected plenty of their values into the future human social stock. In many cases individuals are far less important than now.

- One last fun point: given a highly polished, sophisticated, technological, safe, secure, near-immortal interstellar society, human beings eventually become very homogenous, then rather dull, and eventually bored and suicidal. So, they elect, quite voluntarily, to deliberately inject danger and unpredictable change back into the mix to keep their edge. They reintroduce economic scarcity, money, disease, aggression. In Cordwainer Smith's novels, this becomes "The Rediscovery of Man", and everyone has to choose a newly reinvented "nation" to belong to. Thus there are New French people travelling between the worlds 17000 years from now and drinking Bordeaux in Bistros on the 4th moon of Beta Hydri VIi .

Sorry to digress - hope it was a bit interesting. Whilst I love the Traveller paradigm of SF RPG (which includes Star Wars RPG, Ringworld, and so on), I'm fascinated by the possibilities of a really "out there" SF rpg background!

Now - :focus: (yay - get to use that smiley at last!)

[drohem]

Those are some cool ideas Shaira.

1. I like the concept the human life span being extended so that humans are practically immortal. I like the concept of euthanasia. It could become like Logan's Run where suicide is mandatory at a certain age.

2. Animals uplifted or genetically enhanced to become bi-manual and bi-pedal is a great concept. This has been done in the SF RPG Justifiers. In this future, mega-corporations replaced governments and their scientists have intjected human DNA into animal embryos to create a semi-slave society of anthropomorphic humanoids. Also, if you can get a copy of GURPS Uplift, then you could convert it to BRP. It's a great sourcebook, as all GURPS products.

3. Interstellar space beings that prevent conventional space travel could range from non-intelligent entities to Cthulhu-like creatures. This could inject horror into a space opera setting. Interdimensional travel via folding space or the mind already has a firm place in SF. Robotech and Dune immediately come to mind.

4. Human society becoming homogeneous with a heavy influence from Oriental civilization also has etched its place within SF literature. Blade Runner and Serenity immediately come to mind when thinking about this angle.

You touched upon some really come concepts in science-fiction. Most of this ideas could easily facilitate a campaign style ranging from space fantasy to hard and gritty science fiction.

Good stuff! :thumb:

[camazotz]

I think Greg Stolze's ...In Spaaace! setting is pretty vital to understanding AI.

It's not the AI we need to worry about....it's the Artificial Stupidity that makes 'em pass the Turing Test!

(That aside, neat ideas Shaira, I like the way you think!)

[camazotz]

Put an Ai in a ship and you are limited by space and power.

Put an AI on a planet and you are limited only by the number of components you can put together.

So, I can see planet-based AIs as being a lot more powerful than ship-based ones.

Also, there is the overwhelming Salesman Factor. AIs are made by companies/organisations and are sold by salesmen. Now, salesmen want to charge for absolutely everything they can, so any extra modules will have a cost associated with them. This means that, although AIs could be unlimited, in practical terms you get what you pay for.

For a really good look at just how interesting AI-based ships can get, you should look for Ian M. Bank's novels, especially Excession and it's related novels. Yes, humanity may start with building AI on the planet....but the AI may find that rather, ah, constraining...

[soltakss]

Ian M Banks, yes, I struggled through a couple of those. I found them like swimming through treacle and didn't particularly enjoy them.

The trouble with AIs is that you would have to put so many checks on them as to make them effectively useless.

So, Asimov has his 3 Laws of Robotics. These are fine, except where you have military or combat situations. If you apply them to AIs you can't have AIs with a military capacity. So, they're out straight away.

So you have to put something in so the AI doesn't take charge and knock your own side out. You need a way to countermand an AI or to take it over. But, an evolving AI would probably become immune.

So, you need a way to turn an AI off. But, a distributed AI becomes harder and harder to turn off as its systems become separated. Even if you can turn it off, there's the moral and ethical aspects - if it is Intelligent, is it alive?

It's a minefield, so I prefer my AIs to be small and dumb.

Red Dwarf had the right idea - clever AIs that look stupid so as not to be threatening.

[camazotz]

Ah, well I guess ymmv with Ian M. Banks. Still, I've found him considerably more readable than many current new-wave space opera writers, right behind Charles Stross.

I think a great deal of current SF posits the notion that the grand curve of progress will inevitably lead to runaway AI and a future society in which notions of control are rendered moot, at least for humans, who in turn become the topic of control issues for the dominant AI. Despite some excellent visions of such futures, I still think the "smart yet dumb" AI works best in terms of gaming. The closest any setting book for rpgs has gotten to the topic of advanced AI with an impact is Transhuman Space, and honestly, I felt much the same towards that setting as you did about Mr. Banks' works...