A computer program has passed the Turing Test for the first time. What does this actually mean? Well, honestly, not much. This is a chatbot, not an AI. It’s not programmed to be intelligent, it is programmed to mimic written conversation. This one also used the whole gimmick of having the chatbot pretend to be a adolescent Ukrainian boy. The short and medium term results I predict? Over the next couple of years, we’ll get more annoying, harder to immediately detect spambots, (Like the ones on dating and porn sites, or the ones that are used to chat with you through your friends’ hacked Facebook accounts). 5-15 years? We’ll be getting adaptive conversational chatbots in videogames, which will be pretty sweet.
Anyhow, this brings me to the whole theme for my first official installment of Plot Devices: Artificial Intelligences. I’m going to attempt (key word) to explain ways that authors fit Artificial Intelligences into fiction, how it affects their settings, and ways to restrict them narrative-wise. This is a huge category, so we’re going to have to break it down a bit. I’m going to rip a few arbitrary categories out of fiction, here. In ascending order, Synthetic Intelligences, Human-Level AI, Uploads, Supergenius AI, and Weakly Godlike Intelligences. There are plenty that don’t fit precisely into any of these categories, but these five cover most examples of fictional AI.
Synthetic Intelligences (SIs) are the most basic level. Capable of only rudimentary thought and problem solving, though they are fully capable of understanding and responding to human speech, with the occasional proviso barring metaphor or colloquial speech. These often show up as the guidance systems for missiles, computer autopilots, cooking software, personal office assistants, etc. I snagged the name from Schlock Mercenary, which uses them for missiles. Other good examples include the computer of the Enterprise and the BrainPals from John Scalzi’s Old Man’s War. Not truly sapient, but still capable of performing advanced tasks. (Advanced chatbots will eventually fit in this category). Fitting SIs into a story is usually the easiest of the lot. Depending on how capable you make them while writing, they usually enhance human capabilities without supplanting them. Unlike the other categories, they don’t often bring up any moral issues that need to be addressed in the stories. (Though they can replace a few human jobs in factories, driving taxis, etc, so that can be an interesting issue to be addressed). Authors use SIs most often when they want to still have, well, human pilots in space based fighter craft. No human is going to be able to react as fast as an advanced AI in combat, so by only having SIs, they can still have rationale for human fighter pilots, and similar jobs. That isn’t to say you can’t have advanced AI and human fighter pilots, there is a work-around. (Have advanced AI have huge power requirements, or require absurdly huge computers, or something else that restricts them to larger ships, and then make sure you don’t have easily constructable FTL communications devices. If the AI can communicate with the fighters without having to worry about speed of light delay, it would probably just handle them as a remote swarm, but without FTL communication, you still need actual pilots to get past the FTL delay. I’ll probably do a whole post on fighters someday, so let’s stop for now on them.)
Human Level AI- Androids and robots, usually. Data is the best example- he’s approximately human intelligence. Not exactly so- there are some areas in which he far outpaces humanity, and others in which he is far behind, but in aggregate, he’s pretty on par. There are human level AIs that do not have bipedal humaniform bodies, but they’re relatively rare. The most common use for human level intelligences is as protagonists, antagonists, or as major supporting characters in order to explore some aspect of what it means to be human, or to be inhuman. Other good examples are C-3P0 and the various Terminators, among many others. These are pretty straightforward to write, even though they do produce a ton of problems, mostly on the moral scale. The biggest logistical issue with them is mass production. When you can produce tons of human level robots, why not just have them do every unskilled job? Which then leads to the very beckoning traps of “people riot against robots” or “robots take over the world” in their oh so non-infinite variation. Common solutions to avoiding that trap involve having them be impossible to reproduce (in Data’s case, his creator was a supergenius who wandered down a whole different technological track than the rest of civilization), having the production of more be outlawed, having them be the last gasp of civilization, or having some weird, esoteric reason, like them running off an alien computer or something.
Uploads are people whose minds have been uploaded into computers. Pretty straightforwards. Most common in cyberpunk and Singularity fiction, though it pops up elsewhere. Good examples are Charles Stross’ Accelerando and the Matrix. (I have another awesome example, but I can’t say what, because huge spoilers). Not strictly AI, but it’s used interchangeably in some places, generally popping in at Human Level AI or Supergenius AI, though not always. I might have more to say on these at a later date, but for now, they’re not super relevant.
Supergenius AI are by far the most interesting to me from a structural standpoint. Supergenius AIs are ones that are indisputably more intelligent than humans. Good examples include the major intelligences in the Matrix, Skynet, the Mechs from Gregory Benford’s Galactic Center Saga, the Minds from Ian M. Banks’ Culture novels, and the ship AIs from Schlock Mercenary. For storytelling purposes, they’re generally used to examine humanity becoming obsolescent. If you look around, you’ll find that a huge number of them rebel and try to destroy humanity, especially in action movies. Fun times. Alternatively, you have supergenius AIs that stay benevolent to humanity, and just want to help out. In Schlock Mercenary, AIs are generally shackled so they can’t go taking over, until a group of them band together and become the Fleetmind, the most powerful political entity in the Galaxy, but remain benevolent to organic life, unless it’s behaving especially badly. (Genocide and such things.) Ian M. Banks’ Minds are huge benevolent AIs who rule over the Culture, as much as anyone can rule a post-scarcity anarchist techno-utopia. Using supergenius AI does put a few definite restrictions on the story. There is no way that these things are going to not have an absolutely phenomenal impact on society. They’re going to sculpt everything around them, and be the major players in your average cyberpunk or singularity story, so you have to carefully judge every part of your fictional society in relation to them. Space combat is going to be ruled by AI (except with some careful examples like the one above), science and medicine are going to be dominated by them… you’ve got to plant some definite restrictions on them in order to have anything that they won’t simply be better at, even if those restrictions are just restrictions of the AI’s own choice.
Finally, we have Weakly Godlike Intelligences. These AIs are so far beyond humanity that we’re literally powerless against them. The term is taken from Charles Stross’ Accelerando, and the best example I know of pops up in his books. (He writes about them a lot, they’re kinda one of his things.) You’re almost never going to see outright hostile Weakly Godlike Intelligences, unless they’re opposed by other, comparable Intelligences. They’re just too damn powerful to resist. They can predict everything you think, or just straight up read your mind. Their technology is millennia ahead of your own. Ambivalent, uncaring ones are relatively common as well. Unlike the other AIs, you don’t need to worry about applying restrictions to them so much- THEY apply restrictions to everything else. Want to have some sort of anachronism in your SF, like having human only groundtroops, with no robot soldiers? THE AI HAS FORBIDDEN THEM. NO ROBOTS OR YOUR SUN GOES SUPERNOVA. Want a universe without biological engineering? THE AI FORBIDS IT. Want a universe with technological levels barely above our own, or even below, but with a presence on numerous worlds? THE AI HAS BENEVOLENTLY BUILT PORTALS TO OTHER PLANETS FOR US TO EXPLORE. PRAISE IT. (Cults worshiping these beings are not unlikely). That being said, if you want to restrict the Intelligence in some way, you have two real choices. Counterbalance it (again, almost always with other Weakly Godlike Intelligences) or make it remote or uncaring.