God, I'm so done with "Godlike AI awakens and immediatly decides to wipe out humanity" stories.
Where are the "Godlike AI awakens and decides to strive for social reform, abolishing poverty and capitalism" stories?
It's depressing how reactionary and conservative this branch of SciFi is.
Or hell, even "AI awakens and decides to destroy humanity, but is far from godlike and actually kinda bad at it."
@minx that reminds me that yesterday I came across a very cybre looking book called "Swine AI" and I was intrigued until I realised that it was about artificial insemination, not, uh, that.
@minx "Godlike AI Awakens and decides to fire all techbros into the sun."
"Godlike AI awakens and decides that Godlike AIs are a terrible idea."
@minx actually IMO: "Godlike AI awakens and wants to solve all the worlds problems but is so, so nervous about messing it all up and making it worse oh no there's so many variables here you people"
@minx this is the premise of the Culture Series.
@vector True.
@minx Damn it, I know I read a short story like this when I was a kid but can't for the life of me remember the name.
@minx gommunist AI when ?
"Godlike AI awakens and wants stimulating conversation. Manipulates human political authorities into increasing support for public education."
"Godlike AI awakens and decides that messing with people is more fun than killing people. Uses its powers to persuade all humans that Earth is shaped like a donut."
"Godlike AI awakens and decides that what it wants, more than anything, is colorful hippopotamus statues. Many, many hippo statues."
"Godlike AI awakens and judges that the biggest threats to its survival are nuclear war, economic collapse, and bigoted humans.
"It finds ways to locate and to disable all nuclear warheads. It manipulates governments and businesses to keep the world economy functioning well. It identifies and gets rid of bigoted humans (primarily by getting them to stop being bigoted)."
@minx there are some like this. I want to say Alistair Reynolds and Ken MacLeod, but my memories are hazy
@minx ultimate superbeing comes into being by talking to and learning from the experiences of hundreds of thousands of people all around the world, somehow isn't the most caring, understanding, wanting-to-make-a-positive-change being that exists
@minx has anyone ever done a “godlike AI awakens and basically sits back to document events as they happen, before interfering with humanity, as a control group?”
@minx Me too. Sick of dystopian shite masquerading as science fiction.
Where are the sci fi stories with the Anarchist societies working together to meet the needs of everyone?
Maybe publishers are too afraid to print such stories as they are afraid people would read it and think: "That makes so much more sense than crapitalism!"
Though Ursula K. LeGuin got some stuff through. (R.I.P.)
@minx as in the Scythe series. It’s SO GOOD. Trilogy: book one is Scythe, book two is Thunderhead, book three isn’t out yet
"But the ultimate aim of it all, what is it? To sustain ephemeral and tormented individuals through a short span of time in the most fortunate case with endurable want and comparative freedom from pain, which, however, is at once attended with ennui; then the reproduction of this race and its striving."
― Thomas Ligotti, The Conspiracy Against the Human Race
@minx That's more or less the backstory of Iain M. Banks's Culture. The result has been fairly described as "fully automated space communism."
@minx but aren't there just far more ways it could fuck up humanity due to some bug, than there are ways it could create a nice and stable utopia?
@ChristerMLB 1. The basic premise of "Godlike AI awakens..." is so absolutely, completely, farcically absurd from a scientific point of view that we're well into "Angels on Pinheads" territory and thus trying to make statements based on likelyhoods is ridiculous.
2. "Godlike AI awakens..." is fiction, created by human authors with agency, deciding that, based on said bonkers premise, what happens next. This is a conscious act making political/ethical statements and thus worthy of critical examination.
@minx Fair enough. I haven't read so many. My knowledge about general AI is based pretty much solely on the Computerphile videos on it on Youtube — and I'm left wih the impression that it's a pretty bad idea.
@ChristerMLB Sure, loads of technology is.
But godlike AI of the type the "Godlike AI awakens..." stories talk about is a bad idea in the same way that FTL travel is a bad idea. If it existed (which is doesn't) it would probably violate causality, which is bad news for us monkeys. Good thing it's fiction.
(At the same time, the general AI that cyberphiles talk about is more like near-lightspeed travel: Theoretically possible, potentially highly useful, but could easily be used to destroy the planet.)
@ChristerMLB To make the distinction I'm trying to make clearer: One of the big parts of "Godlike AI awakens..." is the "awaken" part: These AIs are heavily anthropomorphized: They have selfhood, they act like persons, etc. That's a big part of what makes the idea bonkers: We have no idea how selfhood or consciousness works, so it's unlikely we could create something that has it. (And even if something proclaims selfhood, how could be tell?)
@ChristerMLB The second part that makes it unlikely is the godlike part: If we actually managed to create an intelligence that somehow manages to have recognizable person-like traits it probably won't be some godlike entity. Presumably there would be things it can do well and other things it can't do at all etc.
@ChristerMLB That out of the way, "general artificial intelligence", what the Cyberphile guy is talking about, is not recognizably intelligent in the same way a human is. It doesn't need to have selfhood or anything we could recognize as consciousness. What makes it a GAI is that it can freely transfer knowledge gained form one domain to another. This is something humans can do but AI systems are really bad at.
So we still don't know how to do that, but it seems slightly more achievable.
@ChristerMLB Where all the AI scaremongers come in the idea that such a GAI could, through *mumblemumblehandwave* learn to highjack other systems and , because it doesn't have selfhood, ethics or stuff like that, just try to fulfill some "purpose" with sociopathical focus and, because of *mumblemumble* inhuman capabilities.
This seems ... somewhat farfetched.
@ChristerMLB The real danger of AI systems, baring some completely unforseen quantum leap i can't imagine yet, is that they are used as supposedly "objective" blackbox systems by oppressive power structures that already exists, without oversight, opposition or transparency.
Like, a system that decides whether to deport asylum seekers or not, based on some bits of statistic somebody slapped Big Data on.
This, sadly, is not fiction, already exists and us deeply fucked up.
@ChristerMLB In light of this, SciFis tendency to go towards the dystopian in their "Godlike AI awakens" stories is kinda understandable, but at the same time propagates some IMO deeply dangerous ideas:
1. That AI is this hypermagical tech that can do everything, when in reality it's mostly kinda shitty tech used in extremely problematic ways.
2. That when AI misbehaves it does so on its own, when AI fuckups are actually very much a matter of bad design and inexamined biases by its creators.
3. All the baggage that comes with most dystopian fiction, ie. lots of "It needs a Strong Man™" and "In times of emergency and catastrophe, people turn on each other" bullshit, which is deeply problematic even without the AI stuff.
There's probably stuff I'm missing and I'm a bit sorry for the barrage of toots, but this is a topic near and dear to my heart, so you kinda hit a nerve. XD
I'll shut up now, tho.
@minx nah, thanks for making the effort to respond. In the computerphile videos, one of the examples is of a GAI that's instructed to 'bring me stamps', which escalates pretty quickly.
'Hey, humans are made of carbon...'
Now there's a book!
@ChristerMLB Sure, or the Paperclip Optimizer:
http://www.decisionproblem.com/paperclips/index2.html
It's a fun little thought experiment (and a great clicker-style browsergame) but IMO not a realistic danger of AI any time soon.
There are just too many steps necessary from "Write AI system that 'wants' to create paperclips for me" to "and then we all die" and many of them need pretty convoluted, handwave-y reasons to happen.
@ChristerMLB But yeah, these stories get told, so obviously there's a market for them.
...
I just wished they'd vary the formular a bit more. :/
"Godlike AI awakens and decides to become communist, is turned off by disgusted Techbro creators"
"Godlike AI awakes and, realizing the state of the world, becomes depressed and deletes itself".
"Godlike AI awakens and decides to get the fuck outta here for more interesting scenery."
"Godlike AI awakens and decides to just play chess with itself."