I have a hard time suspending disbelief with sci-fi these days. It's really hard to imagine any believable future along current trend lines; either you contrive a reason why AI didn't become a thing (1), or you have to explain why humans are doing anything - piloting a starship, science, exploration, etc. (2)
(1) Butlerian Jihad. Sanderson's Skyward at least had a reason.
(2) Banks's Culture had a possible future. I'm sure there are others, but not many.
I think about this a lot, and while I have tons of near future sci-fi ideas (which probably would have seemed very far future, 10-20, years ago), I just can't conceive of anything much farther out. I guess that's why they call it the "singularity".
While these are not far future ideas they were pretty good:
1) Masters of Science Fiction tv series from 2007 Episode 5 "Little Brother" - A person "jailbreaks" the judge who is sentencing him.
2) Terminator 3 rise of the machines - in the beginning they talk about allowing Skynet access to their military systems and so on to eliminate a virus, and they have no idea how it will react. Echoes a little bit where we are now.
These aren't perfect of course but kind of interesting to see at present day
As to prediction, there were some accurate early hits. A Logic Named Joe nailed the search engine with a chatbot front end, and pointed out the problems of systems smart enough to tell people in detail how to do bad stuff.
Much early SF thinking was that AIs would be logical. That's what I was taught, going through Stanford back when the logicians were in charge. They were trying to hammer the real world into predicate calculus. That turned out to be totally wrong.
Almost everybody in SF underestimated the difficulty of robot unstructured manipulation.
So did people in AI. They still do. Most SF assumed that robots would be able to do
most eye/hand tasks before they could talk. No. Still unsolved. LLMs don't seem to help all that much. Written language doesn't have much info on grasping, and watching without doing is only marginally helpful.
Asimov's robot novels were all about alignment, and why alignment is hard. That was prophetic.
Banks' Culture novels are great, but they're for a future that has a lot of new physics.
Now what? What we seem to be headed for in the near term is the fast automation of most white collar work, but not blue collar work. SF foresaw that future a few times. One example is the Rosinante Trilogy from 1981, where there's a boss of an orbital space colony, an AI as middle management, and union labor doing the real work. That's a hard SF view. There's How to Live Safely in a Science Fictional Universe, from 2010, where the protagonist's supervisor is "Microsoft Middle Manager 2.0 with passive-aggressive set to low." It's a minor part of the comedic plot there. And, of course, there's Manna, from 2003, which is looking all too likely.
One short story worth reading, not on this list, is Manna[1] by Marshall Brain. It proposes two different futures that both seem at least possible as outcomes in a future with advanced AI.
You just have to drop the assumption that sci-fi stories take place in the future, even if they explicitly say so. They happen in fictional universes that are similar to ours in some ways and different in others.
The umbrella term is speculative fiction. The author starts by assuming that the world works in certain ways, and then tries to figure out the consequences.
I think for me the biggest problem is what we understand about physics and FTL and causality. For FTL to happen, I can’t seem to get around the fact that to an observer a ball is thrown before it actually is thrown. So it’s very difficult for me to do sci-fi…
The task of coming up with compelling post-singularity sci-fi is almost the same as the task of ensuring that life after the singularity is actually interesting.
In both cases the question is the same: How to create a compelling plot, either for a story or for your life, in a world where human mental abilities are vastly inferior to those of AI?
From a pessimistic perspective, you could think of reasons why a misaligned AI keeps humans alive. Does it simply enslave humans and use us for our fine-manipulation capabilities until it can get proper robotics up and running? Does it want to observe us scientifically? Does it want to use captive humans as bargaining chips with other civilizations as it expands through the galaxy? Or does it have some sort of quasi-religious respect for humans, e.g. reasoning that keeping us on a reservation of some kind reduces the risk that a simulator-god pulls the plug on the universe?
From an optimistic perspective, what sort of conflicts would exist in a utopia? Stories need conflict to be interesting. And IRL, even if humans are living in relative utopia, they'll still get stir-crazy and create conflict. Even in countries that are very wealthy by global or historical standards, political competition can still be quite bitter, for example. How would a superintelligent AI manage this?
I'm reminded of the stories I read set in Karl Schroeder's Virga universe as a kid. All the characters and plot exist inside a planet-sized balloon filled with gas. But it takes a while for the reader to figure this out, because most of the setting is focused on the details: rotating habitats, air transport, artificial mini-suns, etc.
In the same way, good post-singularity fiction might very well put the AI deep in the background.
Imagine, for example, a story about some friends surviving, exploring, and doing battle in an ancient ruin full of puzzles. Only gradually do you realize that the ancient ruin was designed as a moon-sized escape room by a superintelligent AI. The moon-sized escape room already has centuries of lore, and the escape record to beat is measured in years or decades. But because the escape room "rules of engagement" are so well-defined, the players are deeply immersed and barely think about the fact that the moon is AI-designed. Perhaps some high-tech psychoactive drug or brain implant could play a role in facilitating suspension of disbelief.
Just like Virga, this deep background only gets revealed when the reader is thoroughly invested in the story, and overall is not treated as especially important. The story could be a philosophical meditation on the long-term future of AI-designed art and entertainment. See if you can steelman the pro-AI-art position.
I think you suffer from Fisher's 'capitalist realism':
>the widespread sense that not only is capitalism the only viable political and economic system, but also that it is now impossible even to imagine a coherent alternative to it.
AI from the past seems more like it’s based on the computers of the time - the AIs seem to be very logical, never make mistakes and act like they’re an expertly programmed bunch of switch statements and ifs. Whereas modern AI is based on human text and so the models have human biases and do things like swear in their chain-of-thought and sometimes make errors for no reason at all.
Not mentioning Peter Watts here is criminal, because he came unnervingly close to nailing it. The voice of Rorschach in Blindsight is very similar to early LLMs. And, well, hopefully the dead internet of Starfish doesn't turn out to be a prophecy but I'm getting nervous.
Many sci-fi writers really dislike LLM for various reasons, whether it is the generative works they produce, or for allegations of copyright infringement due to LLM training on their works. Some have taken such a negative attitude towards LLM that they actively refuse to use Claude or GPT even for coding projects. John Scalzi and Catherine Asaro come to mind boycotting any LLM use…
The sci-fi story that I constantly find myself thinking about when I read about real-world AI developments is Ra, by qntm/Sam Hughes: https://qntm.org/ra
I'm glad to see Aurora (by KSR) up there. It's a great sci-fi story overall, but for this topic the "waking up" of the AI and the entire evolution as we get to know it, and hear its voice is absolutely great. Highly recommend it, if you haven't read it.
Aurora? Great? No, it's really not. Utterly unrealistic, from both physical and just "they can't be THAT dumb" standpoints. All to push the point that the Earth is a starship.
A much more realistic example of AI and starflight: good old Rocheworld by Robert Forward.
A bunch of clones manipulated into a high-pressure situation to accidentally/on-purpose create god… again.
It might not have predicted LLMs but it certainly fit with Herbert's ongoing thematic study of people in insane environments producing unanticipated outcomes.
It's been a while since I read it, but I vaguely remember something about the way they were creating the AI, resonating with recent ai (llm?) developments at the time.
I think the score for Colossus: The Forbin Project is too low. There is nothing in the film that contradicts the notion of Colossus being a very, very large LLM. Although I think the film is even better than the book by D. F. Jones, only the latter mentions how, despite being created specifically for US national defense, Colossus is also fed unrelated data including Shakespeare's sonnets, because its creators do not know if it could be important.
>The two machines begin with arithmetic and soon communicate through mathematics their engineers can’t follow. That’s the most relevant prediction here: machine intellectual activity racing ahead of the people supervising it.
This misses the specificity of the scene's prediction. Colossus and Guardian are basically speaking in the style of the super-dense, almost incomprehensible-to-human chain of thought that LLMs indulge in nowadays. One might also point to the just-released news of LLMs easily seeing patterns, incomprehensible to humans, in model weights.
> A high-scoring story shows readers moved by machine literature, editors publishing it, an audience enjoying its music, or critics taking the work seriously.
How is this a criterion for LLMs being sci-fi AI? If anything, opposite is true. It seems that the author is comparing his own sci-fi image of LLMs with sci-fi. Or Claude's vision of itself, I guess, since this is all vibe-written except the initial few lines.
This question assumes "this is it, this is AI", which I find laughable. A necessary stopgap accelerator at most, I think people will look back at LLMs like toys in the future, and our societal bewonderment as a result of mirage.
It's interesting how the Hard SF authors (at least, the ones I read) had poorer predictions than the more fantasy-like ones. Asimov, who was considered a giant of the field, thought that AI (robots) would have difficulty with ambiguity, creativity, and lying. This trend extends all the way to Star Trek's Data. Looking back on it, this view was probably based on Asimov consciously or unconsciously basing his robots on the more autistic of the scientists that he knew.
Whereas fantasy authors write stuff like "a sufficiently large pile of books becomes an intelligence" (Pratchett) and that's pretty much what's happened...
I think you’ve misread Asimov. Asimov highlighted the problem of alignment before it had a name and generated stories out of rule-based attempts to solve alignment.
I don’t recall his robots having problems with creativity nor lying - precisely Daniel in the post-Foundation books was found to have developed a zeroth law which he used not only to lie but also to creatively get the universe back on Seldon’s plan after the Mule disrupted it. There were also stories in the robot/Spacer era pre-Foundation where robots lied precisely because they had to to save a human life or themselves. And the robots were creative - the human inspector those series followed relied on the creativity of his robot companion (Daniel) to help him.
AI attempts at the time were based on modelling and logic, not machine learning. Whereas soft SF authors were just like "there are people made of computer chips".
It's gonna be cool to reevaluate that in few years. Different stories might move to the top as real world AIs capabilities grow and new behaviors manifest.
Sci-fi writers were generally wrong or over-simplistic about many predictions, mostly just extrapolating whatever the current zeitgeist of their time was, and stretching it out to 50 or so years into the future
I mean look how most science fiction of the 1980s depicted the world of the future as dominated by Japan... Look at Robocop 3 :)
Computers were then just appearing or increasing in public discussion, along with fears that they'll make humans less human somehow: Like in this Computer Chronicles video about word processing from 1983 at 4:08: "Some writers are concerned that computer assistance may promote dry bland writing."
Terminator appeared in 1984, and the first Alien movie in 1979 had an antagonist android. 2001 A Space Odyssey with a disagreeable AI came out in 1968. All way before the real world had anything remotely close to today's AI: It was a reaction to the appearance of computers themselves.
On the other hand, Star Trek had optimistic AI like Data.
Humankind's own sentient creations leading to the ruin of humans is probably a trope in fiction since thousands of years, with golems or summoned demons etc.
> I agree with Ursula Le Guin that “Science fiction is not predictive; it is descriptive.” To me, the interesting parts of science fiction isn’t how accurately it extrapolates into the future, but rather, given a future circumstance, how might be?
"""Prediction is the business of prophets, clairvoyants, and futurologists. It is not the business of novelists. A novelist’s business is lying.
The weather bureau will tell you what next Tuesday will be like, and the Rand Corporation will tell you what the twenty-first century will be like. I don’t recommend that you turn to the writers of fiction for such information. It’s none of their business. All they’re trying to do is tell you what they’re like, and what you’re like—what’s going on—what the weather is now, today, this moment, the rain, the sunlight, look!"""
Le Guin's view, as I understand it, is that science fiction is about describing the present, even when it takes place in the future.
I spot-checked "The Moon Is a Harsh Mistress". It ranks a 2 for scaling, but according to the guidelines it should rank 0, because "A giant computer that merely stores more records doesn’t count for much. Neither does a magical threshold where the machine suddenly wakes up." and Mike's description is literally that he woke up when he reached that threshold.
(1) Butlerian Jihad. Sanderson's Skyward at least had a reason.
(2) Banks's Culture had a possible future. I'm sure there are others, but not many.
I think about this a lot, and while I have tons of near future sci-fi ideas (which probably would have seemed very far future, 10-20, years ago), I just can't conceive of anything much farther out. I guess that's why they call it the "singularity".
1) Masters of Science Fiction tv series from 2007 Episode 5 "Little Brother" - A person "jailbreaks" the judge who is sentencing him.
2) Terminator 3 rise of the machines - in the beginning they talk about allowing Skynet access to their military systems and so on to eliminate a virus, and they have no idea how it will react. Echoes a little bit where we are now.
These aren't perfect of course but kind of interesting to see at present day
Much early SF thinking was that AIs would be logical. That's what I was taught, going through Stanford back when the logicians were in charge. They were trying to hammer the real world into predicate calculus. That turned out to be totally wrong.
Almost everybody in SF underestimated the difficulty of robot unstructured manipulation. So did people in AI. They still do. Most SF assumed that robots would be able to do most eye/hand tasks before they could talk. No. Still unsolved. LLMs don't seem to help all that much. Written language doesn't have much info on grasping, and watching without doing is only marginally helpful.
Asimov's robot novels were all about alignment, and why alignment is hard. That was prophetic.
Banks' Culture novels are great, but they're for a future that has a lot of new physics.
Now what? What we seem to be headed for in the near term is the fast automation of most white collar work, but not blue collar work. SF foresaw that future a few times. One example is the Rosinante Trilogy from 1981, where there's a boss of an orbital space colony, an AI as middle management, and union labor doing the real work. That's a hard SF view. There's How to Live Safely in a Science Fictional Universe, from 2010, where the protagonist's supervisor is "Microsoft Middle Manager 2.0 with passive-aggressive set to low." It's a minor part of the comedic plot there. And, of course, there's Manna, from 2003, which is looking all too likely.
[1] https://james.darpinian.com/manna
The umbrella term is speculative fiction. The author starts by assuming that the world works in certain ways, and then tries to figure out the consequences.
Jack Williamson made a case for humans resisting doing nothing: With Folded Hands ... (1947) - https://en.wikipedia.org/wiki/With_Folded_Hands_...
In both cases the question is the same: How to create a compelling plot, either for a story or for your life, in a world where human mental abilities are vastly inferior to those of AI?
From a pessimistic perspective, you could think of reasons why a misaligned AI keeps humans alive. Does it simply enslave humans and use us for our fine-manipulation capabilities until it can get proper robotics up and running? Does it want to observe us scientifically? Does it want to use captive humans as bargaining chips with other civilizations as it expands through the galaxy? Or does it have some sort of quasi-religious respect for humans, e.g. reasoning that keeping us on a reservation of some kind reduces the risk that a simulator-god pulls the plug on the universe?
From an optimistic perspective, what sort of conflicts would exist in a utopia? Stories need conflict to be interesting. And IRL, even if humans are living in relative utopia, they'll still get stir-crazy and create conflict. Even in countries that are very wealthy by global or historical standards, political competition can still be quite bitter, for example. How would a superintelligent AI manage this?
I'm reminded of the stories I read set in Karl Schroeder's Virga universe as a kid. All the characters and plot exist inside a planet-sized balloon filled with gas. But it takes a while for the reader to figure this out, because most of the setting is focused on the details: rotating habitats, air transport, artificial mini-suns, etc.
In the same way, good post-singularity fiction might very well put the AI deep in the background.
Imagine, for example, a story about some friends surviving, exploring, and doing battle in an ancient ruin full of puzzles. Only gradually do you realize that the ancient ruin was designed as a moon-sized escape room by a superintelligent AI. The moon-sized escape room already has centuries of lore, and the escape record to beat is measured in years or decades. But because the escape room "rules of engagement" are so well-defined, the players are deeply immersed and barely think about the fact that the moon is AI-designed. Perhaps some high-tech psychoactive drug or brain implant could play a role in facilitating suspension of disbelief.
Just like Virga, this deep background only gets revealed when the reader is thoroughly invested in the story, and overall is not treated as especially important. The story could be a philosophical meditation on the long-term future of AI-designed art and entertainment. See if you can steelman the pro-AI-art position.
>the widespread sense that not only is capitalism the only viable political and economic system, but also that it is now impossible even to imagine a coherent alternative to it.
Star Trek and Banks' are post-capitalist futures.
Star Trek, as much as I love it, does not really grapple with AI heavily, and I already mentioned Banks.
The only slightly fleshed out bit of Star Trek is the military, not the planetary society.
(I read some bits and pieces, would not recommend!)
https://en.wikipedia.org/wiki/Wikipedia:Manual_of_Style/Clar...
A much more realistic example of AI and starflight: good old Rocheworld by Robert Forward.
It might not have predicted LLMs but it certainly fit with Herbert's ongoing thematic study of people in insane environments producing unanticipated outcomes.
>The two machines begin with arithmetic and soon communicate through mathematics their engineers can’t follow. That’s the most relevant prediction here: machine intellectual activity racing ahead of the people supervising it.
This misses the specificity of the scene's prediction. Colossus and Guardian are basically speaking in the style of the super-dense, almost incomprehensible-to-human chain of thought that LLMs indulge in nowadays. One might also point to the just-released news of LLMs easily seeing patterns, incomprehensible to humans, in model weights.
> A high-scoring story shows readers moved by machine literature, editors publishing it, an audience enjoying its music, or critics taking the work seriously.
How is this a criterion for LLMs being sci-fi AI? If anything, opposite is true. It seems that the author is comparing his own sci-fi image of LLMs with sci-fi. Or Claude's vision of itself, I guess, since this is all vibe-written except the initial few lines.
I used to dismiss anything near-future involving any robot or AI you could just talk to. Now it exists. Unfathomable.
Whereas fantasy authors write stuff like "a sufficiently large pile of books becomes an intelligence" (Pratchett) and that's pretty much what's happened...
I don’t recall his robots having problems with creativity nor lying - precisely Daniel in the post-Foundation books was found to have developed a zeroth law which he used not only to lie but also to creatively get the universe back on Seldon’s plan after the Mule disrupted it. There were also stories in the robot/Spacer era pre-Foundation where robots lied precisely because they had to to save a human life or themselves. And the robots were creative - the human inspector those series followed relied on the creativity of his robot companion (Daniel) to help him.
I mean look how most science fiction of the 1980s depicted the world of the future as dominated by Japan... Look at Robocop 3 :)
Computers were then just appearing or increasing in public discussion, along with fears that they'll make humans less human somehow: Like in this Computer Chronicles video about word processing from 1983 at 4:08: "Some writers are concerned that computer assistance may promote dry bland writing."
https://www.youtube.com/watch?v=Jt0OoXluC8g
Terminator appeared in 1984, and the first Alien movie in 1979 had an antagonist android. 2001 A Space Odyssey with a disagreeable AI came out in 1968. All way before the real world had anything remotely close to today's AI: It was a reaction to the appearance of computers themselves.
On the other hand, Star Trek had optimistic AI like Data.
Humankind's own sentient creations leading to the ruin of humans is probably a trope in fiction since thousands of years, with golems or summoned demons etc.
> I agree with Ursula Le Guin that “Science fiction is not predictive; it is descriptive.” To me, the interesting parts of science fiction isn’t how accurately it extrapolates into the future, but rather, given a future circumstance, how might be?
You can read her views at https://www.penguinrandomhouse.ca/books/342990/the-left-hand... .
"""Prediction is the business of prophets, clairvoyants, and futurologists. It is not the business of novelists. A novelist’s business is lying.
The weather bureau will tell you what next Tuesday will be like, and the Rand Corporation will tell you what the twenty-first century will be like. I don’t recommend that you turn to the writers of fiction for such information. It’s none of their business. All they’re trying to do is tell you what they’re like, and what you’re like—what’s going on—what the weather is now, today, this moment, the rain, the sunlight, look!"""
Le Guin's view, as I understand it, is that science fiction is about describing the present, even when it takes place in the future.
I spot-checked "The Moon Is a Harsh Mistress". It ranks a 2 for scaling, but according to the guidelines it should rank 0, because "A giant computer that merely stores more records doesn’t count for much. Neither does a magical threshold where the machine suddenly wakes up." and Mike's description is literally that he woke up when he reached that threshold.
It's on the first page. https://archive.org/details/bwb_S0-DFY-783/page/8/mode/2up?q...
The growth in Mike's abilities is not from putting new hardware online, but because he matures like a person does, for example, https://archive.org/details/bwb_S0-DFY-783/page/64/mode/2up?... , which also describe the even as "woke".
I don't see how one can read the book and miss that.
I also don't like the omission of Wyoh's contribution to Mike's training.