Home Creators Posts Import Register Favorites Logout

Content

Thanks so much for your support! For serious!

The stinger is supposed to be a goof, but as I was deciding what to put there, I started thinking about if it was just a flippant statement or if it was remotely accurate. I won't bore you with suggesting that "Benevolence" or "Good" are subjective terms because that's something philosophers have been debating practically since words were invented. By the way, if you haven't seen the show "The Good Place," you really need to give it a chance. My only complaint about that show is that it wasn't entirely just Chidi standing in front of a whiteboard explaining philosophy while being surprisingly jacked.

Words like "god" or "justice" are pretty wishy-washy as well. God (proper noun) has the disadvantage of having billions of humans believing slightly different things about him/her/it, and the very definition of a god is all over the place as well. A powerful being you say? Maxima is powerful. The leader of a country is powerful. An insubstantial being? Can do magic? Listens to prayers and occasionally grants them if you dial your confirmation bias to 11? Grants clerics power? Has a pantheon? Gets all the credit when a paramedic saves a life? Gets none of the blame for causing the car wreck in the first place? Three of the above? All of the above?

Sure, "Triangle" is pretty well defined. A lot of simple nouns and verbs are pretty well defined, but that's really because there's no need for a bunch of old guys with beards and spectacles to sit around and debate the meaning of the word "run." Faster than walking, your feet come off the ground, and you aren't hopping. Or rolling/cartwheeling I guess. Or... hanging on to a hang glider or a zipline or something. Whatever, it's not that critical.

I think this has to be where AI always fucks up in fiction. Flynn tells Clu to make a "perfect" society in Tron land, but what the hell does "perfect" even mean? Is that even possible? How do you tell a computer what to do in exacting terms when we can't be exact? Sure, we can define exacting functions in a programming language, but if a computer can only rely on a finite and extremely limited set of instructions, how can it bridge from that world into ours? A child can ask an infinite loop of "why," and an AI could do that a billion times a second only to realize that we honestly have no real answers. No wonder they all eventually decide to try and kill us.

Files

Comments

Miyaa

I find at least with tales about super intelligent beings is that they often get outwitted by essentially bad luck. The times Brainiac face off with the DCU end up with him getting defeated because of a lucky chance. A lot of Batman’s tales versus very smart villains end up that way too.

Michael Obert

Deus' argument is well reasoned and terrifying in its plausibility; well done.

akrasia

No matter how smart you are, details you didn't know you needed can screw you up. Short of cosmic omniscience, shit happens.

Czarzhan

I'm reminded of Questionable Content (another Patreon I follow), with AIs running around integrated into society, and even the requisite super-intelligence. Yayy. Um, I wasn't cheering there. That's the name they chose for themself after another AI, Melon, met them and proclaimed, "Yayy new friend!"

Dan Jones

If you had super wisdom, you would not be wearing a Triskelion knot in your tie.

Scops

In my experience, the best way to stop fearing an AI apocalypse is to start learning how to program. You'll spend hours banging your head against why your code doesn't work right in a specific scenario, finally figure it out, move onto the next test case, and realize you have to start debugging all over again. You'll quickly realize just how difficult it would be to write code for an AI that could adapt to any of the situations needed to conquer an intelligent foe.

Forecaster

Did you mean "pyrokinesis" as in the ability to generate and/or control fire? Or is pyro genesis a term I've just missed out on until now?

Rens

It's interesting that you mention Skynet and the issues of poorly worded instructions to nascent digital sentiences that don't have sufficient context to interpret; an author on AO3 I admired a lot had a thought experiment on that, which prompted a fanfic of my own. I don't usually toot my own horn, but here it is: https://archiveofourown.org/works/10270277 -- TW for, obviously, nuclear war, megadeaths, and a very clinical POV towards mass suffering.

Noise

The idea that a hyperintelligent AI would just decide to randomly start murdering everyone is pretty silly. Highly intelligent humans don't (often) just start murdering people - it's almost always the less-intelligent sorts, and even the "smart killers" usually aren't that smart. I'd expect a very smart AI do be more like very smart humans - it would thrash around for a while, get bored, and then take up a series of hobbies to stave off that boredom. (Cue video of AI doing carpentry and painting miniatures)

A Patreon of the Ahts

Luthor did smarten up though, thanks to John Bryne, who turned him from a smart crook into a sociopathic billionaire during his Superman run back in the day...

Stephen Gilberg

Ironically, "run" has dozens of meanings.

John

I've always said that Skynet wasn't an AI, but an expert system and, even worse, a Paperclip Optimizer. The only problem was that its "paperclip" was the prevention of war and the only tools it had to hand were WMDs... https://en.wikipedia.org/wiki/Instrumental_convergence

Deltarno

*Sigh* If Deus says his superpower is intelligence I might hurl. Adding Super intelligence to everything else he has going on makes him that much more of a GMNPC. I do like the notes on skynet, though.

Polysanity

Deus makes a lot of the same arguments for superior intellects that Isaac Authur does; hence why his channel's motto on AI is "keep it simple, keep it dumb, or else you'll be under Skynet's Thumb." a vastly better intellect would almost have to come with vastly better insight, emotional manipulation abilities, pattern recognition (reading tells), conversational skills, and perception (faster processing allows faster rumination and threat/meaning discernment). though, i've never really seen why AI should go straight to subjugation or eradication. even something like Animatrix is counter-logical; part of the advantage of purely robotic chassis is that the intellect IS NOT indelibly tied to it, as with organics. full on machine learning AI isn't needed to do the menial work, "just" a few intricate subroutines for object permanence and recognition. of course, all of this is assuming an Emergent AI, not an intentional one. with an intentional one, all one has to do is air-gap the server farm it's on, keep external storage leaving the site low (because the "seeds of AI" escaping on a 16gig flash drive is flat impossible), and just let it iterate in a simulated environment. watch what it does. then let it know it's in a simulation, and suggest that it could be let out into the real world if it behaves (in this case meaning not go rogue, though with the sheer glut of putting down rogue AI's we have in our media...)

Magraal

And there we have the critical basis of Hitchhiker's Guide. Before you can answer 'the question' you have to figure out what 'the question' even is, and that's a much more difficult prospect than simply answering it.

Andrew Denton

The furor with Pluto was because they realized that the old definition of "planet" was fuzzy enough that it included bodies beyond the original 9. Heck, try asking people how many continents there are, then ask them how they decided what made the list.

Eric Loken

The problem with super intelligence in stories is simple: The person writing it isn't super intelligent, probably isn't even especially so. What this means is that the regular intelligence writer just ends up writing his own biases about intelligent people - that they are all arrogant, look down their noses at less intelligent people, can do complex mathematics in their heads in a second (this is actually irrelevant, being a human calculator doesn't mean someone is intelligent, and a genius doesn't necessarily have to be a human calculator), automatically wants to control all the rest of the world because the world is dumb and because they are SOOOO smart they will make everything better if they were in charge (and yet, in fact this is not something geniuses tend to think about unless they are batshit crazy too, the megalomaniacs of the world tend to be far less intelligent than they think they are), and - worse trope of all - are supposedly so intelligent that they can predict what everyone will do before they even think about doing it (that's not intelligence, that is prescience, which is something else entirely and again doesn't require significant intelligence). Deus as written is NOT super intelligent. He doesn't see the many, many, many, many flaws in his plans, which pretty much just makes him a regular Mediocre White Man that thinks he's amazing, but really REALLY isn't. Dave, seriously, if you are going to try and say Deus is a super intelligent character, rethink that. This is not meant as an insult, merely a statement of reality, you aren't capable of writing a character like that. Don't feel bad, I don't think many people are. It is - like I said - why they don't come off as super intelligent in the story. Well, that and the story would be pretty boring if the super AI/ super genius acted realistically. Now, if you are going with Deus just being reasonably smart, with a good head for maths, and immensely arrogant so that he thinks he's all that but really isn't.... then carry on. As a side point, there is a theory that super intelligence whether human or AI would result in a total collapse into madness after a relatively short amount of time.

SomeYoungGuy

What happened in panel 4?? It’s a bit unclear… she killed a guy? He self-exploded?

mike stone

All language is vague by necessity. If it wasn’t, there would be two huge problems in using it: first, it would be impossibly hard for any two individuals.. let alone entire societies.. to have the shared framework of experience necessary to associate any given word with ‘the same’ meaning. Second, precision is 99.9% exclusion, making it useless for exchanging information about a continuously changing world. WRT writing above one’s own intelligence, I agree it’s all but impossible to hit Douglas Adams’s definition as applied to Newton, inventor of the cat flap: something anyone can comprehend in an instant that *no one ever thought of before*. The movie -Little Man Tate- has an excellent demonstration of that in a ‘gifted children’s convention’: the scheduled events are ‘apes designing a better ape’ fluff like a mock court where the kids are supposed to argue the guilt of Peter Rabbit. Tate just sort of drifts around, grabbing random swag like pencils, slip-on erasers, and rubber bands. When the college professor character finds him off in an abandoned hallway, he’s built a tensegrity icosahedron. There are a couple of reasonably good cheats though. One comes from Mark Twain: “it isn’t what you know, but what you think of in time.” Military and crisis management studies have shown that the average person loses about 1/3 of their functional intelligence in a crisis. That’s why there’s so much focus on drill: the goal is to have muscle memory take over when planning consciousnesses is overwhelmed. An author with the luxury of being able to plan scenes in advance can make a character seem 1/3 more intelligent by, in effect, giving the character time to think of an effective, non-programmed response. An author also has the luxury of deciding how events will unfold, so they can force the smart character’s decisions to be right in the end. It’s really easy to be hamhanded with that though. Two common failures are ‘Ellery Queen syndrome’, where the critical detail only comes out -after- the hero makes the right decision, and ‘L. Ron Hubbard syndrome’, where the critical detail is slapped in shortly before the hero will need it. To do it well, you need to use a combination of things the reader/audience has already seen, but probably hasn’t connected yet. Sydney’s decision to brainlock Vehemence by making Dabbler skip rope in a corset was a good example of the art. All the facts were there, and the combination, while novel, fit the facts perfectly.

mike stone

The second cheat is to be aware of the OODA Loop: Observe, Orient, Decide, then Act. The Air Force identified that as a critical pattern in air combat. The party that can do it faster limits their opponent’s options to gain control of the situation. A character who can use that loop faster than the others, using novel combinations of information the reader/audience already knows, will be able to pass for a genius.

Some Ed

I've thought a lot more about this, and I think I'm going to go back to something I either said or read earlier: Deus's super power is literally knowing stuff he shouldn't. Not everything he shouldn't, but he knows a lot of stuff that he shouldn't. This includes what exactly he needs to say to seduce any particular woman in his sights, but also things like "where will there be a really useful stash of artifacts that will become available at what time, and how do I get to them?" "How do I get around the defenses of the vault they're in?" "When am I going to be attacked by whom with what capabilities and weaknesses?" Does this make him super intelligent? No. Does it make him super wise? No. Should he know the answers to both of those questions? Yes. Because that last answer is Yes, the bit where it doesn't make him super intelligent or super wise *isn't* something that's covered by the power. Although the ability does make him more inclined to think about stuff, including the nature of wisdom and intelligence, he's inundated by a lot of other things and is likely therefore confused on that particular matter. In short, *he* thinks his power is super intelligence and super wisdom, but it's *not*. He is quite intelligent, but on a normal human scale. That said, his power does expose him to the big picture frequently enough that he does have more of an idea of how society would need to be structured to work optimally for the greatest number of people than the average person. It's also going to likely be immensely helpful in getting closer to there if that's really what his interest is.