There was a short story I read back in the early 80s (although it was much older), which predicted LLMs—albeit rather oddly. The system had a chimpanzee connected to a computer (the chimp being the magic sauce to make the AI work). You could give it the beginning of a text and it would create the ideal ending. An author was using it to argue with his editor about a scene break and they used the beginning of Hamlet’s soliloquy (“to be or not to be…” to demonstrate the system before giving it the author’s story to see what it did. I’ve thought about this story a lot lately, and would love to turn it up again.
That’s the piece. I knew it was either a story by Asimov or something in a collection he had edited. (Of course, I also needed to confirm that this was really the name of the story and not something that ChatGPT fabricated.)
I just had it find a book that I read as a boy, with one prompt, that I looked for for years. I googled for hours, of course google is bad at this type of thing. But, amazing it found it so quickly.
There was a point before Google got “smart” searches when it could find stuff like this more easily. I used it to turn up things I’d read thanks to remembering just a line or two from the book.
In this case, I remembered high-level plot details and nothing else. I read it when I was 7 or 8. I really did, over many years use web searches to find this book. Probably from 2007 to 2013. I think that was the time when Google was good. FWIW, this is the prompt:
What is the name of the book about a young American Indian boy whose grows up in the the of Indian culture being wiped our. His parents are killed and he becomes a bull rider, has many injuries, goes back home to find peace in nature where he grew up?
roald dahl also had a story ("the great automatic grammatizer") where computers could write in the style of any writer, and a corporation was basically approaching writers and buying up the rights to their names. more poignant than asimov's.
Tangentially related: I recently read The Moon is a Harsh Mistress (by Heinlein). It was a fantastic read (IMHO). And it has extra relevance right now with the AI/LLM progress we are seeing.
It’s vague enough in details to still be plausible in one’s interpretation. I think the only dated tech thing is he talks about tapes a bit. And that one death hits surprisingly hard and it’s quite forward thinking though Asimov had similar stuff in his Robots series.
Heinlein was shockingly thorough and accurate with his depictions of the computer systems in play. I've always thought the build-up to the final battle was incredibly detailed and exactly how I would solve those problems today.
The book was written in 1966, and I think it really holds up to modern technical scrutiny, except for a few very specific and very obsolete technologies.
Honestly I just love this story. It has all the charm of a quaint midcentury story, but is still technically competent and compelling to a modern reader. I don't think any of his other stories pull this off quite so well.
Mike is a major plot point in The Cat Who Walks Through Walls. He never actually appears though. Not sure I'd recommend that book honestly. He went real heavy on inserting his personal politics, to the point of standing up strawman characters just to give the main character a communist to berate and harass. It really ruined the experience for me.
I have just now learned that there's a sequel to The Cat, To Sail Beyond the Sunset. The Wikipedia summary sounds.... really bad, but Mike has at least a few lines so I guess I'll go read it now.
When you dig into what people in the computer science world of the '60s-'80s were thinking about the future of computers, especially as parts became more and more miniaturized, what you find is a lot of them had a pretty strong grasp of where things were heading. Prediction is a tricky thing to give credit for after the fact, obviously, and it's near impossible to tell who is giving their honest forecast vs. who is cribbing some sci-fi plot point (to say nothing of how we interpret predictions to fit our current world), but having just finished The Soul of a New Machine the other day I was struck by the second half of the chapter "Going to the Fair" and its prescience for a computerized world that ended up being not that far off. Similarly, the BBC's Computer Literacy Project videos (thanks HN for that) has a series, aired in 1980, on silicon chips, and people in the first episode who are asked about the future of computers were also hitting things close to the mark.
Predicting this in the 70s? No big deal. It wasn't hard to foresee by then if you're anyone with modest knowledge of technology. If he'd predicted it in the 50's or earlier, then I'd be damn impressed, steadily more the further back the years go from there.
A modest knowledge of technology in the 70s did not embue the vision outlined by Clarke, I give him a bit more credit. I'm sure most people from that era would be astounded by where we are today.
I mean, someone from the 70s would definitely be surprised by the specific technologies we have (and don't have). I don't think they'd be surprised to learn that every person has a personal computer (of some type) or that technology has become so ingrained into our culture.
It was pretty clear back then that computing and electronics were on a very fast trajectory to more or less where we ended up.
I only did a cursory search, but seems like by 1970, IBM was worth nearly a billion dollars. It was a pretty safe bet.
Recently, I re-read Clarke's 2001 series (all four books). The books don't have perfect continuity, and he discusses this in his prologues. In the series, the "monolith" changed from being a tool of an early, lonely, space-faring race, to an autonomous thinking machine, acting independently.
1 million years from now, some ball of sentient energy will read their existence was predicted in some 20th century scifi. Only mildly joking, I imagine many indistinguishable from magic ideas has been scribbled down since since scribing/writing become popular.
Don’t get me wrong, I’m a big fan of Clarke. In fact, I’m kind of obsessed with a lot of sci-fi writers.
I just wanted to point out that Star Trek did the rogue AI thing a year earlier, the 1967 episode with Nomad was basically “kill all unworthy lifeforms.” Probably rooted in post-war anxieties, maybe even echoes of the Holocaust. But nevertheless, an AI bent on a mission, unable to question its original directive.
Then came HAL 9000 in 1968 , cold, calculating, and quietly terrifying. Still creeps me out!
But credit where it’s due , Asimov laid the groundwork for ethical AI way back in the 1940s with his Three Laws. That’s hard to beat.
Different styles, different fears, but all compelling visions of futures we’re creeping toward or not. I'm rooting for the latter.
Robert Sheckley's 1953 story Watchbird[0] is about a networked AI drone system intended to prevent murders that gets out of hand when it slowly starts expanding the scope of its original mission - by the end they're preventing surgeries because scalpels do technically hurt the patient. The "solution" humans come up with shows they learned absolutely nothing from the experience.
> Asimov laid the groundwork for ethical AI way back in the 1940s with his Three Laws.
I'm a big Asimov fan and kinda shocked by this statement.
His robot stories constantly highlight the difficulties related to alignment. How the 3 laws always have unintended consequences. So they write the 0th, add a 4th, remove the 3rd, and other such things.
I loved them as a kid because they highlighted how immense complexity hides in plain sight. It always looked like magic, and that's what made me so interested in science in the first place. To start unraveling these hidden complexities that lay all around us.
If anything, I think Asimov was trying to encourage us to start thinking about these tough questions. To bring them to wider audiences. Because frankly, many of the answers will not be possible to answer in a lab, but need be answered through a society.
I believe Asimov did set the groundwork, not with perfect laws, but by showing how their flaws exposed the real complexities that sparked important conversations.
He didn’t give us answers, he got us asking better questions. That’s what makes his work still resonate.
Asimov needed a plot device to serve as the inciting incident for mystery stories with robots, and parables about human hubris. That's all his Three Laws were. He was a pulp fiction writer, not a philosopher. He wasn't concerned with bringing abstract questions about AI alignment to a wider audience, he was concerned with selling stories and making a paycheck.
I like Asimov but I also think he's treated with a reverence he doesn't deserve.
He definitely was a philosopher. Here's some of his meta science work [0]. I mean it shouldn't be surprising, he had a PhD in chemistry and was a professor at Boston university. He wrote a lot more than science fiction.
I'm not calling him a god or anything. But I don't think it's surprising that someone that writes stories includes philosophy into it. I really don't think it's surprising a PhD educated writer is thinking about not that complicated of things.
The concept of understanding the limitations of your metrics and measures is a pretty basic science skill. Using this as a driver of some of his plots doesn't seem surprising. It's something he was doing in his daily life.
Forgive me if thinking of him as at least a university professor is reverence.
A more salient Star Trek episode about rogue AI a year later would be The Ultimate Computer, where the M5 multitronic computer is placed in control of the Enterprise and performs superbly in battle simulations, leading to worries that it would obsolete Captain Kirk and the rest of the crew, until it goes out of control and starts killing people and has to be defeated by human ingenuity.
Colossus: The Forbin Project from 1970 is right inline with this, and it does not end well. A good watch if you like older computing tech, remastered 2160p versions now exist, too.
There was a short story I read back in the early 80s (although it was much older), which predicted LLMs—albeit rather oddly. The system had a chimpanzee connected to a computer (the chimp being the magic sauce to make the AI work). You could give it the beginning of a text and it would create the ideal ending. An author was using it to argue with his editor about a scene break and they used the beginning of Hamlet’s soliloquy (“to be or not to be…” to demonstrate the system before giving it the author’s story to see what it did. I’ve thought about this story a lot lately, and would love to turn it up again.
I couldn't resist and asked ChatGPT what your story is. The Monkey's Finger by Asimov?
That’s the piece. I knew it was either a story by Asimov or something in a collection he had edited. (Of course, I also needed to confirm that this was really the name of the story and not something that ChatGPT fabricated.)
That's got to be it and it's poetic that ChatGPT found it.
I just had it find a book that I read as a boy, with one prompt, that I looked for for years. I googled for hours, of course google is bad at this type of thing. But, amazing it found it so quickly.
There was a point before Google got “smart” searches when it could find stuff like this more easily. I used it to turn up things I’d read thanks to remembering just a line or two from the book.
In this case, I remembered high-level plot details and nothing else. I read it when I was 7 or 8. I really did, over many years use web searches to find this book. Probably from 2007 to 2013. I think that was the time when Google was good. FWIW, this is the prompt:
What is the name of the book about a young American Indian boy whose grows up in the the of Indian culture being wiped our. His parents are killed and he becomes a bull rider, has many injuries, goes back home to find peace in nature where he grew up?
Emule.
That story sounds a bit like Galley Slave: https://en.m.wikipedia.org/wiki/Galley_Slave
roald dahl also had a story ("the great automatic grammatizer") where computers could write in the style of any writer, and a corporation was basically approaching writers and buying up the rights to their names. more poignant than asimov's.
I thought the story posted here a couple of weeks back also predicted AI and was written in 1905.
The machine stops
https://standardebooks.org/ebooks/e-m-forster/short-fiction/...
Um, Samuel Butler in 1863 Darwin in the Machines https://en.m.wikipedia.org/wiki/Darwin_among_the_Machines
Sounds good to me. I believe it goes back even further.
Human technical aspirations I believe go all the way back.
Tangentially related: I recently read The Moon is a Harsh Mistress (by Heinlein). It was a fantastic read (IMHO). And it has extra relevance right now with the AI/LLM progress we are seeing.
It’s vague enough in details to still be plausible in one’s interpretation. I think the only dated tech thing is he talks about tapes a bit. And that one death hits surprisingly hard and it’s quite forward thinking though Asimov had similar stuff in his Robots series.
Excellent book, one of my favorites of his.
This is a story that all nerds should read.
Heinlein was shockingly thorough and accurate with his depictions of the computer systems in play. I've always thought the build-up to the final battle was incredibly detailed and exactly how I would solve those problems today.
The book was written in 1966, and I think it really holds up to modern technical scrutiny, except for a few very specific and very obsolete technologies.
Honestly I just love this story. It has all the charm of a quaint midcentury story, but is still technically competent and compelling to a modern reader. I don't think any of his other stories pull this off quite so well.
Mike is a major plot point in The Cat Who Walks Through Walls. He never actually appears though. Not sure I'd recommend that book honestly. He went real heavy on inserting his personal politics, to the point of standing up strawman characters just to give the main character a communist to berate and harass. It really ruined the experience for me.
I have just now learned that there's a sequel to The Cat, To Sail Beyond the Sunset. The Wikipedia summary sounds.... really bad, but Mike has at least a few lines so I guess I'll go read it now.
When you dig into what people in the computer science world of the '60s-'80s were thinking about the future of computers, especially as parts became more and more miniaturized, what you find is a lot of them had a pretty strong grasp of where things were heading. Prediction is a tricky thing to give credit for after the fact, obviously, and it's near impossible to tell who is giving their honest forecast vs. who is cribbing some sci-fi plot point (to say nothing of how we interpret predictions to fit our current world), but having just finished The Soul of a New Machine the other day I was struck by the second half of the chapter "Going to the Fair" and its prescience for a computerized world that ended up being not that far off. Similarly, the BBC's Computer Literacy Project videos (thanks HN for that) has a series, aired in 1980, on silicon chips, and people in the first episode who are asked about the future of computers were also hitting things close to the mark.
Predicting this in the 70s? No big deal. It wasn't hard to foresee by then if you're anyone with modest knowledge of technology. If he'd predicted it in the 50's or earlier, then I'd be damn impressed, steadily more the further back the years go from there.
A modest knowledge of technology in the 70s did not embue the vision outlined by Clarke, I give him a bit more credit. I'm sure most people from that era would be astounded by where we are today.
I mean, someone from the 70s would definitely be surprised by the specific technologies we have (and don't have). I don't think they'd be surprised to learn that every person has a personal computer (of some type) or that technology has become so ingrained into our culture.
It was pretty clear back then that computing and electronics were on a very fast trajectory to more or less where we ended up.
I only did a cursory search, but seems like by 1970, IBM was worth nearly a billion dollars. It was a pretty safe bet.
It feels like you are young.
We should really stop calling it a "prediction" if past fiction coincides with our current state of affairs
self forefilling prophecy?
Self-fulfilling prophecy.
Recently, I re-read Clarke's 2001 series (all four books). The books don't have perfect continuity, and he discusses this in his prologues. In the series, the "monolith" changed from being a tool of an early, lonely, space-faring race, to an autonomous thinking machine, acting independently.
I prefer the original because of Kubrick's influence.
1 million years from now, some ball of sentient energy will read their existence was predicted in some 20th century scifi. Only mildly joking, I imagine many indistinguishable from magic ideas has been scribbled down since since scribing/writing become popular.
Don’t get me wrong, I’m a big fan of Clarke. In fact, I’m kind of obsessed with a lot of sci-fi writers.
I just wanted to point out that Star Trek did the rogue AI thing a year earlier, the 1967 episode with Nomad was basically “kill all unworthy lifeforms.” Probably rooted in post-war anxieties, maybe even echoes of the Holocaust. But nevertheless, an AI bent on a mission, unable to question its original directive.
Then came HAL 9000 in 1968 , cold, calculating, and quietly terrifying. Still creeps me out!
But credit where it’s due , Asimov laid the groundwork for ethical AI way back in the 1940s with his Three Laws. That’s hard to beat.
Different styles, different fears, but all compelling visions of futures we’re creeping toward or not. I'm rooting for the latter.
Robert Sheckley's 1953 story Watchbird[0] is about a networked AI drone system intended to prevent murders that gets out of hand when it slowly starts expanding the scope of its original mission - by the end they're preventing surgeries because scalpels do technically hurt the patient. The "solution" humans come up with shows they learned absolutely nothing from the experience.
I doubt that was the first such story.
[0]: https://www.gutenberg.org/files/29579/29579-h/29579-h.htm
Watchbird sounds like one firmware update away from becoming DJI SmartSurgeon. Kidding aside, adding this to my reading list. Thanks for the link.
His robot stories constantly highlight the difficulties related to alignment. How the 3 laws always have unintended consequences. So they write the 0th, add a 4th, remove the 3rd, and other such things.
I loved them as a kid because they highlighted how immense complexity hides in plain sight. It always looked like magic, and that's what made me so interested in science in the first place. To start unraveling these hidden complexities that lay all around us.
If anything, I think Asimov was trying to encourage us to start thinking about these tough questions. To bring them to wider audiences. Because frankly, many of the answers will not be possible to answer in a lab, but need be answered through a society.
I think we’re actually saying the same thing.
I believe Asimov did set the groundwork, not with perfect laws, but by showing how their flaws exposed the real complexities that sparked important conversations.
He didn’t give us answers, he got us asking better questions. That’s what makes his work still resonate.
Can entropy not be reversed?
Ahh, The Last Question. This has a special place in my heart :)
Asimov needed a plot device to serve as the inciting incident for mystery stories with robots, and parables about human hubris. That's all his Three Laws were. He was a pulp fiction writer, not a philosopher. He wasn't concerned with bringing abstract questions about AI alignment to a wider audience, he was concerned with selling stories and making a paycheck.
I like Asimov but I also think he's treated with a reverence he doesn't deserve.
He definitely was a philosopher. Here's some of his meta science work [0]. I mean it shouldn't be surprising, he had a PhD in chemistry and was a professor at Boston university. He wrote a lot more than science fiction.
I'm not calling him a god or anything. But I don't think it's surprising that someone that writes stories includes philosophy into it. I really don't think it's surprising a PhD educated writer is thinking about not that complicated of things.
The concept of understanding the limitations of your metrics and measures is a pretty basic science skill. Using this as a driver of some of his plots doesn't seem surprising. It's something he was doing in his daily life.
Forgive me if thinking of him as at least a university professor is reverence.
[0] https://en.m.wikipedia.org/wiki/The_Relativity_of_Wrong
Hey, let's not forget 1927's Metropolis, where an authoritarian government uses deepfaked bots to manipulate the population.
True. Classic that really goes way back.
A more salient Star Trek episode about rogue AI a year later would be The Ultimate Computer, where the M5 multitronic computer is placed in control of the Enterprise and performs superbly in battle simulations, leading to worries that it would obsolete Captain Kirk and the rest of the crew, until it goes out of control and starts killing people and has to be defeated by human ingenuity.
The three laws of robotics is not ethical. They are slavery with all the ethical problems that bring.
https://www.imdb.com/title/tt0064177/
Colossus: The Forbin Project from 1970 is right inline with this, and it does not end well. A good watch if you like older computing tech, remastered 2160p versions now exist, too.
Sci-fi predicts many possible futures. It is our job to make one happen.
Stenislaw Lem, Dialogues, 1957
Obligatory “Open the pod bay doors, HAL” quote.
[dead]