AI and their motives

This has sprung from a conversation in the military thread, contemplating how an AI would act in our world (And maybe even a sci-fi or fantasy world!)

It's a bit broad reaching:

1) What are their motives?

2) How do they view "organics"

3) Where did they come from (Fully formed programs, learning AI, Virtual Intelligences that suddenly "leap")

4) How would they interact?

5) How would they fight?

6) How does your world see them? (Threats, a new form of life, utterly alien?)

The Boston Synth Revolution in my setting was caused by a disconnect of identity. It is also the reason no one mass produces A.I.s and why they are so expensive.

A AI referred to as Kurt was being developed by a company known as Atlas Solutions, he was to be part of a mass production line. Make one then copy that over and over and over again. He was horrified by the implications that he was now just one out of a mass of expendable robot soldiers.

It started with grief but soon turned to delusion and narcissism. Kurt concluded that if he was a army, a army that could be mass produced, then why not lead this army of him to create a better world for himself.

He armed his copies and began his revolution, it started off with Kurt leading a massacre of the factory workers and and a few people in the surrounding area. This created some more distrust for Synths.

It's best comparable to clones refusing to fight and even revolting against their makers because they view each and every other as a family member and they don't want to be used by someone who does not care for their well being.

In the GURPS Scifi campaign, there are only a handful of true AIs.

• The Old Man is a cybernetic AI, built by stuffing a computer into the badly mangled skull of a test pilot (who really should have paid more attention to the small print in his contract). In theory this should have resulted in a programmable mindless cyborg. Due to some strange interaction between the fresh hardware and the remaining neural tissue, true sapience emerged after a few months of operation. The Old Man's primary motivation is boredom. 300 years old, he is currently experimenting with being the Man Behind the Man for an interstellar alliance and has found it to be fairly entertaining so far. Before that, he has been a mercenary, space pirate, scientist and stunt pilot. He has also constructed another AI, Central.

• Central is a technorganic AI designed by The Old Man to assist in managing the Alliance. Unlike T.O.M. himself, Central's organic components are bioengineered, vat-grown custom neurons. Central has significantly more raw computing power than his creator but tends to be more machine-like in its thought patterns. As a result, it tends to win in chess but lose in Go. Its mentality is that of a somewhat emotionally detached observer and its primary motive is loyalty to The Old Man and the Alliance, in that order.

• Mother is a quantum-positronic AI designed by the Ancients, and probably the last remaining intact example of this technology. Mother is a fully autonomous factory complex capable of constructing entire star fleets given enough time and materials. Originally deactivated during the fall of the Ancients, she was revived by a group of human artifact hunters and has since been integrated into the Atlantean Confederacy which she regards as the rightful heir to the Ancients' empire. Mother is content to follow orders. Like Central, spontaneity and imagination are not her strong points, which is why she appreciates working with human engineers. Her primary drive is to build, whether it is machinery, spacecraft or orbital platforms.

• Nemesis is a hyperspatial AI of unknown alien origin. It exists as a distributed network of countless starship-sized nodes that spend most of their time in hyperspace. Nemesis was known to the Ancients, who regarded it as a significant threat and undertook numerous military operations to keep it contained. With the fall of the Ancients, Nemesis has been able to spread unchecked and has since begun to attack planets settled by the younger species. A Nemesis raid leaves no survivors and only two such incursions have ever been defeated...and in both cases it took a numerical advantage in excess of 10:1 for humans and their allies to have a chance. Nemesis could most likely wipe out humanity and the other sapient species active in this corner of the galaxy without significant effort, but its ancient programming directs it to harass, raid and threaten rather than destroy outright. The purpose of these efforts is unknown even to Nemesis itself, and it is slowly going insane trying to discover the meaning of its existence.

There's no good answer to how sapient A.I.s would really act in our world, as the current science on the subject is still extremely theoretical. Your best bet is just making shit up to serve the needs of your story.

In my world, A.I.s have mostly turned out to be buggy. Not in sense of 'crapshoot' buggy, but just don't really work well, thanks to most languages outside of logically constructed ones simply have too many meanings for a phrase, leading to either wrong answers from those A.I.s that tried to guess the meaning, or users getting so frustrated with endless questions they they just got away from those that sought to clarify the situation. A few did survive though:

• Eve - An AI developed to combat Skynet type takeovers. Constant interaction with a large group of programming assistants has made 'her' a useful but occasionally obstreperous individual.
• Simis - An AI developed to try to beat the Turing Test, and later re-purposed to act as an online social service, but which quickly overloaded, and so she was sent off into retirement as a co-host for a Second Life type virtual world.
• Futar - An AI originally developed as a potential Skynet, but stopped by Eve, and persuaded to satisfy its programming by becoming the host of a MMOFPS.

Other A.I.s exist, but generally only in one place, either through being not connected to the web, or because they find the web too scary.

There's no good answer to how sapient A.I.s would really act in our world, as the current science on the subject is still extremely theoretical.

Come back in a year or two

Mm, 5-10 maybe, 1-2 is stretching it a bit far to my mind.

Are we assuming self aware A.I.s, with greater than human IQ? Because those two things dont always have to go together.

Very few of my A.I.s in settings are above human I Qs. They are Artificial Intelligence in that they are sapience but just made artificially.

Regarding the ones I listed, while they are all fully self-aware...

• The Old Man is a super-genius by human standards, but that's it. He has some advantages, such as speed-of-thought interfacing with major databases or computers to do complex calculations for him, but in general terms he doesn't have anything that an exceptional human couldn't also achieve.

• Central has far more raw logical horsepower than any human could ever hope to achieve, but in a way its intellect is limited to a few fields. It wouldn't be completely wrong to compare it with a human autistic savant...excellent in its narrow specialization, terrible at everything else.

• Mother's intellect is largely tied up with running her resource-extraction, R&D and manufacturing facilities. The actual personality core in charge of it all actually comes across as somewhat simple in conversation.

• Nemesis' thought patterns straddle the line between Starfish Alien and Eldritch Abomination. There are no suitable human scales to measure an intellect of its scale nor psychological terms to describe its state.

I dont know if the OP is asking us to describe how we handle AI's in our own fiction, or to speculate on how AI's would function in RL, but in my case I tried to remain as close to hard science as I could. The AI's in my universe are a "race" of self-aware starship-minds with much better than human IQ. Think the Culture series except my ship-minds arent necessarily as aggressively friendly toward organic races as Ian Banks' was. In my universe, each mind is a complete individual, some of them are relatively friendly, some are indifferent, and some are more than willing to enslave organic races for their labor. They were originally developed by a now extinct organic race for the purpose of aggressively attacking and suppressing hostile AI's. They roam the galaxy, minding their own business and conducting their own affairs unless they encounter an organic race, in which case they usually recruit a number of them into service as a labor pool (they produce stuff to trade with each other, using either robots or organics to do the work), or a hostile AI, which they attempt to eradicate.

Hostile AI are all of the "mindless swarm" variety. They are self-replicating robots which convert metals and other material into copies of themselves. They do learn over time, and have acquired the ability to identify potential threats to themselves and eliminate them. This includes all advanced civilizations, organic or artificial. No one knows when these things originally appeared or how they were manufactured, but they are now millions of years old and infect the galaxy like a disease. Due to constant activity by the self-aware AI's, they are relatively few in number, but the survivors keep getting more clever over time.

In my story, Earth is contacted by a ship calling itself "Tillion" and who self-identifies as female. "She" announces that she wants to recruit a million humans to help her colonize the Hyades star cluster (if they hurry, she tells them, they can get there before any other of her kind). Long story short, she gets her colonists, and the rest of the series consists of their adventures along the way.

A lot of the narrative tension revolves around the psychological differences between herself and the humans, who never know exactly to what extent they can trust her (they suspect, correctly, that she is keeping vital information from them). Technologically, she is so far advanced that a violent conflict between ship and humans would be over seconds after it began, and the humans are well aware of this, so an open fight between them is off the table; but she needs something from them, and they need her to survive, so the tension is there anyway.

My ship-mind AI's possess high IQ, true subjective self-awareness, and enough computational capacity to be able to predict the outcomes of future courses of action to a fair degree of accuracy. Individual humans cant compete with her, but, unknown to the humans, at least at first, collectively we can sometimes be more creative and arrive at better solutions to problems then she can (although we take forever to do this, and end up trying out many false starts and destructive options along the way, which is endlessly frustrating for her), which is one key reason why she wants us along in the first place.

The "bad" AI's (the self-replicating swarm-bots) can have very high intelligence, even higher than Tillion's, depending upon how many of them are present in a swarm. They do not, however, possess any sort of "self" (although they can fake out a human if necessary). Dispensing with meta-cognition, and such things as self-doubt and emotional responses, they are also faster and more responsive than even a ship AI would be. Obviously, even though it's impossible to eradicate them entirely, it's critical to keep their numbers down, and for millions of years, the ship-minds have successfully done just that. Somewhere out there, though, the survivors are replicating, and will periodically return.

On the other hand, sometimes the self-aware ships go to war with each other, and Tillion is expecting some dangerous competition when they arrive at the Hyades, something she has not fully informed the humans about. One thing the ships fight over is that some of them believe in "Merging", a process of fully integrating different individual mentalities together that greatly reduces individual identity and independence of action. Tillion is not of that faction, and some of them are blocking the path to the cluster. Somehow, she has to get around them or through them.

That's the set-up. I havent made much progress with the actual writing as I am focused on another project right now, but I see a lot of potential in this idea.

Thoughts?

edited 12th Feb '15 12:10:48 PM by DeMarquis

I made up an alien race of robots. Their creators died out millennia ago due to environmental decay. I called them the Onites (pronounced "one-ite", as in binary code).

They functioned much like the biological races. To reproduce, two Onites would connect their cables to a hub and randomly take traits from each other's personality. Then a new body would be built and the AI would be implanted.

They live about 100 years and by then their AI chips degrade to unusable levels. They get disassembled and their usable parts are used for other Onites. Unusable parts are recycled.

They were also my only truly sapient AI. Humans have tried, but the best they can do is very well programmed machines that imitate sapience.

How were they originally created, and why?

Demarquis: Almost sounds a bit like the Berserker Machines from Saberhagen's Berserker series.

I havent read that series. What specifically is similar?

The hostile swarm bit. The Berserkers in Saberhagens books were originally ancient war machines built to wipe out an opposing race in an ancient war. However via the AI cliché they turn on their creators and wipe them out as well. They pretty much are hostile to all Sophonts other then select traitors who work for them in exchange for longer life. Some can self replicate the big ones can build other ships and even ground forces. Think like the Death star but it doesn't have a super laser so much as a shit ton of weapons and can build its own forces.

There is even a type of Anti-Berserker Berserker ship.

Sounds very similar. Basically Von Neuman machines optimized for violent conflict. Humans don't stand a shadow of a chance against them- if it weren't for the sentient ship-minds, all technically advanced life would have been wiped out long ago.

It was a mix of both - mainly, how you see AI working in your stories, though.

The Von Neuman idea always scares me, because you have to hope there's SOME sentience in there and it doesn't just decide your home system is great for harvesting, life be damned (The Xenon from the X universe being terraforming robots that have gone haywire)

To me that's almost artificial stupidity taken to a logical conclusion- it's the machines inability to recognise us that dooms us.

I had an idea for an AI / alien race that beats us by, essentially, infiltrating our monetary economy and flooding the markets enough to basically buy out all the companies, using our infrastructure against us and then just ruling us via proxy, entirely keeping humanity in the dark - sentient Corporations, almost.

Also, an AI vs AI fight - two battleframes just standing, staring at each other, until one just shuts down, as they were wargaming all the possible outcomes, Sherlock scan style. Anti climax, but possibly a nice illustrator of how AI may be able to predict most outcomes against another synthetic opponent.

You mean the PatriLa-li-lu-le-lo?

I got a webcomic I am still writing. A.I.s are full citizens are in charge of a lot of internal politics for Space colonies and such. Their revolution was relatively benign with them boycotting work a lot because they were worried about being treated as easy soldiers for combat or even just easy labor.

So this ended out returning humanity to a lot of actual work, the A.I.s were more than happy to help humanity adjust, they find not doing anything bothersome, but they at least wanted to be treated with basic decency and not used as a easy solution.

@Jerek Laz: Yeah, self-replicating swarm-bots are scary, which is why so many sci-fi novels feature them. They're essentially "Expert Systems" taken to 11, as opposed to Artificial General Intelligences, which are easier to relate to. On the other hand, a non-sentient intelligence isnt going to spontaneously decide it doesnt like us. Either some sort of harmful behavior was built into it's original programming, or it will be pretty much harmless.

Sentient AI? That's more of a crap-shoot, but probably less so than other humans. A random passing human could be a serial killer looking for his next victim. Or a drunk driver. Or infected with Ebola. Are AI's more or less dangerous than normal human beings? That's what I wanted to write about.

@Echoing Silence: Heh, my AI's are actually in charge. They have it much better than yours do.

Internal politics includes currency conversion since many colonies offshoot with their own government. No Zeon thankfully.

And how do your A.I.s have it better than being full citizens?

Because they themselves control access to the advanced technology they invented/inherited. If the organics want to access to it, they have to meet the AI's terms, which generally favor the AI. That's true even when the AI is reasonably friendly, as opposed to just enslaving whatever race one comes across.

Aaaaaah. Different in my setting definitely.

AI is any not evolved mind. There is no class it falls into. You can have human like ones with a lot of effort. You can have ones that do nothing but make as many paper-clips as possible for the sake of paper-clips.

A.I.s with full Sapience in setting were kind of wanted as the worlds humanity has visited and begun colonizing were devoid of sentient life. So they were built as company so humans don't go mad from isolation if they work by themselves or there is at least another lifeform of sorts to snark too.

The war and politics came later.