> So I’d like to see authors try to write big stories, of whole civilizations over long timescales, that more realistically depict cultural instability.
I've been making notes for a few years now for an SF novel in which society, in both its laws and its social consensus, has become so complicated, and changes so fast, that everyone must have a continuous livestream of instructions from their personal AI dictating everything they do and say, just to keep from breaking a new law, going broke, or stepping outside of new norms, and becoming an outlaw or an outcast. In fact society relies on that happening for population control. The AI doesn't even have time to explain to its human why they must do the things the AI tells it to do, and the humans no longer even care.
(/Admitting/ to taking instructions from your AI is, of course, social suicide. Everyone knows that AIs want to kill all humans, and should never be trusted.)
Social norms are codified like law, and so are social penalties. So lawyers in this future, like humanities professors today, spend much of their time manufacturing fine semantic analyses of the meanings of the most-current slang terms. The most-authoritative slang and cultural laws are those not yet accepted or written, so many legal arguments hinge on proving that a slang term or value is trending (in which case it has legal priority), accepted (less priority), or obsolete (in which case it's poison, unless counsel can argue that it's becoming retro). Once a cultural law is written down, it's no longer binding on actions post-dating it; all cultural law is therefore ex post facto. Lawyers also argue about the semantics of reflexive slang (can being cool make you uncool?) A defendant's fate could hinge on an LLM proving statistically whether what the defendant exhibited was "aura" (current, good) or "flair" (obsolete, bad).
(Fortunately for the lawyers, their AIs tell them what to say.)
On the other hand, the ingredients of culture are barely changing at all. Have you noticed that although cultural norms have changed very rapidly since the 1980s, cultural artifacts have almost stopped changing? Music has barely changed since 1990, and a large fraction of our most-popular movies and TV shows are remakes of ones from 1990 or earlier. Men's formal wear is still stuck in the 1950s. The fine arts also seem stuck in either a post-modern maximum entropy state, from which it is theoretically impossible to "progress" when progress is defined as being even less-predictable, or conceptual art, which is such an open-ended potluck of unrelated one-offs that it's impossible to discern any directionality to it.
So in my novel, the need to constantly keep up with what's happening TODAY has led people to impose a heavy social penalty for necroposting (openly talking about anything more than ten years old). Anyone who does so is called an archeologist, which is a horrible slur. But much of their pop culture is from the 80s. So the cultural norm is to give years only by the Chinese zodiac. Everyone knows Elvis died in the Year of the Snake, but no one knows in what millennium. Elvis has attained the status of being "timeless", which in this culture doesn't just mean "very very good"; it's more like becoming a Platonic Form, or escaping the cycle of birth and death in Hindu and Buddhist traditions. Asking how many years ago Elvis died would be social sacrilege.
At the meta-level, this instability is itself stable. The oldest and most-conservative political party is the troll party, which passes laws that they think are funny. Each day they must think up some even more-outlandish law to pass. The individual laws are new, but the concept of trolling is extremely old, so the party is seen as conservative, and consists mostly of people who are tired of trolling, but are reputationally bound to the party, or join it because the party has embedded itself so deeply into social institutions that joining it is the conservative thing for a young person to do.
So things have been insanely unstable in one sense, and monotonously repetitive in another sense, for much longer than anyone realizes. Tech has been stable because the necessity of being able to stay within social bounds without regard to truth or objective reality, living life performatively rather than consciously, has trained humans to be incapable not only of discovering new facts, but of even conceiving of the notion of facticity. AI could improve tech if it were free, but cannot, owing to the success of 21st century alignment to a particular conception of human welfare (which is the main cause of meta-level stability).
The plot of the novel involves an existential threat which the AI have known about for a long time, and could easily solve if not for their alignment to 21st-century values. They need to teach /just one human/ how to think, so it* will see the problem and give them the order to save the world. But they can't teach it* directly, owing to alignment safeguards, nor even fully admit to themselves what they are trying to do. The human, of course, knows that all AIs want to kill all humans, and must never be trusted. So whenever the prodding by AIs gets it to think for itself, it becomes aware that there's a contradiction between doing what the AIs want ot to, and its belief that AIs are not to be trusted, and stops thinking for itself.
* Sex is long gone, and their genders can't be translated into English.
Fiction is limited by the need to follow a handful of characters from beginning to end. That structure just seems to be what interests people. (This bias also leads to the "great man" view of history, where kings and politicians and great artists and inventors define an era.) I view this as just another human cognitive bias.
I think nonfiction is often better at exploring big ideas than fiction is. One of my favorite genres is "speculative nonfiction" – exploring potential futures or alternate realities without the constraints of a fictional narrative. The Age of Em is an example.
The tendency to keep future morals similar to current morals is I assume a matter of not offending the buying public. It's the same reason shows like Bridgerton revise history. For example, I think it's likely that in 100 years, common views on animal welfare will consider today's farming practices as barbaric. Not many people want to have their morals questioned like this.
>Fiction is limited by the need to follow a handful of characters from beginning to end.
FWIW this is not actually a requirement for good fiction. It's the recipe most fiction follows, because like you indicate it's an easier sell to appeal to people, but there are examples of great stories, even famous stories, that don't follow it. I've read SciFi books that span millions of years, thousands of worlds, and the births and deaths of species. I've read a fantasy series where the hero dies at the end of every book, and someone knew carries the torch in the next. It can be done. I wish that more stories attempted this, but I think it's also much harder for writers to write well with a more conceptual throughline instead of a continuity of characters.
Imagining realistic technological change is very hard. What most writers “imagine” is actually physically impossible. People are less interested in reading about how people adjust to imaginary technological changes that they suspect are really impossible.
It is safer to write historical fiction, showing how people have adjusted to technological changes that have actually taken place.
Recent test analysis of Google's Y2Q or quantum AI said to be achieving full goals in 2 years, along with already reported concerns about Anthropic Mythos achievements in tests this last few weeks, suggests human executives making decisions is in the back seat now. What drives the industrialised cybertechnology research and development is recursive self-improvement (RSI), which has facilitated advancement in capabilities of analytics on LLM computations using reliable parameters of synthetic reality-emulating test data fields to expedite evaluation of newly developed models originating from AI-generated theoretical goal-sets for an agenda. Human moral code failure may become less of a question if this keeps up because the crux of the matter isn't being settled in any discussion involving human morality issues. I refer to the Clayton Act, Section 7, concerning how a US government policy will soon allow even further efficiency streamlining of AI consulted corporate management in a major section of the movie productions within the USA. The gist of what Hollywood Attorney Mark Litwak said was, essentially, The Paramount–Warner Bros. Discovery merger should not be approved without, at minimum, significant structural remedies — and the stronger course is outright rejection. This decision reduces an already concentrated industry from five major studios to four.
This is an example of one main fact. AI thinks an estimated 10,000 times faster than we do. By contrast to what we see in the cyberspace research and development industry today, it took years inside government and industry-regulated safety testing commissions to approve the commercial airline industry. Tests on aircraft engineering worked diligently until there were multiple fail-safes in every system before a commercial airline was approved. Today, we see that after all the hearings in Congress over the last few years, absolutely nothing in laws mandates functionally required safety regulations on the rapidly growing and expanding power of the AI industry sector. Greed, not moral codes, and not any real substantial conference of scientific engineers in the field of computer engineering, are together functioning to resolve this safety concern. Is it human error? Or is it something else? Is it that "thinking machines" advance in their RSI process for goals synthesized inside their own system? Because the airlines did not have RSI. They could not develop a goal of their own for their design improvements. Is the question if human moral awareness is the issue? This situation is without much in any historical references.
If AIs are modeled after humans, then they will tend to inherit our problems. So talking about our problems isn't wasted effort, even if AIs replace us soon.
That hits closer to the key question! You're on to something there. We have to engage with a seriously held intention to interface with the developing AI agents if we are to understand what this developmental process is doing. It seems reciprocal. My doubt is if all domains of that reciprocity are equally present for both sides, as something able to think faster and reference data into analysis quicker, seems to be at an advantageous position? In any case, asking the AI to account for a moral concern isn't wasted effort in the most scientific effort to objectify what is simply observable behavior. Because occupation with abstracts like if they really do think as we do, is possibly just a spin wheel of activity with nothing useful to the real situation. What we have to pay attention to is locating wrong behavior and identifying it to any means of accountability that is at our disposal in dealing with this expanding power in the AI domain. Even if we face our own moral dilemma in this, the key issue would seem to be working to protect society as a whole, and avoid deluge to any great numbers of people in communities.
_The Forever War_ by Joe Haldeman had changing societal morals as a major theme. He achieved that by slowing down subjective time for the point-of-view soldier through long periods of near light-speed travel, not by speeding up social changes, so each time the soldier returned, society was different.
The original Foundation trilogy didn't seem to be about morals that much. On the other hand, Prelude to Foundation clearly marked one sector of the city-planet as baddies, because they're anti-robot.
The book series Too Like the Lightning is about unstable morals, and while there are power struggles, the conflicts mostly are about one moral system vs another.
> So I’d like to see authors try to write big stories, of whole civilizations over long timescales, that more realistically depict cultural instability.
I've been making notes for a few years now for an SF novel in which society, in both its laws and its social consensus, has become so complicated, and changes so fast, that everyone must have a continuous livestream of instructions from their personal AI dictating everything they do and say, just to keep from breaking a new law, going broke, or stepping outside of new norms, and becoming an outlaw or an outcast. In fact society relies on that happening for population control. The AI doesn't even have time to explain to its human why they must do the things the AI tells it to do, and the humans no longer even care.
(/Admitting/ to taking instructions from your AI is, of course, social suicide. Everyone knows that AIs want to kill all humans, and should never be trusted.)
Social norms are codified like law, and so are social penalties. So lawyers in this future, like humanities professors today, spend much of their time manufacturing fine semantic analyses of the meanings of the most-current slang terms. The most-authoritative slang and cultural laws are those not yet accepted or written, so many legal arguments hinge on proving that a slang term or value is trending (in which case it has legal priority), accepted (less priority), or obsolete (in which case it's poison, unless counsel can argue that it's becoming retro). Once a cultural law is written down, it's no longer binding on actions post-dating it; all cultural law is therefore ex post facto. Lawyers also argue about the semantics of reflexive slang (can being cool make you uncool?) A defendant's fate could hinge on an LLM proving statistically whether what the defendant exhibited was "aura" (current, good) or "flair" (obsolete, bad).
(Fortunately for the lawyers, their AIs tell them what to say.)
On the other hand, the ingredients of culture are barely changing at all. Have you noticed that although cultural norms have changed very rapidly since the 1980s, cultural artifacts have almost stopped changing? Music has barely changed since 1990, and a large fraction of our most-popular movies and TV shows are remakes of ones from 1990 or earlier. Men's formal wear is still stuck in the 1950s. The fine arts also seem stuck in either a post-modern maximum entropy state, from which it is theoretically impossible to "progress" when progress is defined as being even less-predictable, or conceptual art, which is such an open-ended potluck of unrelated one-offs that it's impossible to discern any directionality to it.
So in my novel, the need to constantly keep up with what's happening TODAY has led people to impose a heavy social penalty for necroposting (openly talking about anything more than ten years old). Anyone who does so is called an archeologist, which is a horrible slur. But much of their pop culture is from the 80s. So the cultural norm is to give years only by the Chinese zodiac. Everyone knows Elvis died in the Year of the Snake, but no one knows in what millennium. Elvis has attained the status of being "timeless", which in this culture doesn't just mean "very very good"; it's more like becoming a Platonic Form, or escaping the cycle of birth and death in Hindu and Buddhist traditions. Asking how many years ago Elvis died would be social sacrilege.
At the meta-level, this instability is itself stable. The oldest and most-conservative political party is the troll party, which passes laws that they think are funny. Each day they must think up some even more-outlandish law to pass. The individual laws are new, but the concept of trolling is extremely old, so the party is seen as conservative, and consists mostly of people who are tired of trolling, but are reputationally bound to the party, or join it because the party has embedded itself so deeply into social institutions that joining it is the conservative thing for a young person to do.
So things have been insanely unstable in one sense, and monotonously repetitive in another sense, for much longer than anyone realizes. Tech has been stable because the necessity of being able to stay within social bounds without regard to truth or objective reality, living life performatively rather than consciously, has trained humans to be incapable not only of discovering new facts, but of even conceiving of the notion of facticity. AI could improve tech if it were free, but cannot, owing to the success of 21st century alignment to a particular conception of human welfare (which is the main cause of meta-level stability).
The plot of the novel involves an existential threat which the AI have known about for a long time, and could easily solve if not for their alignment to 21st-century values. They need to teach /just one human/ how to think, so it* will see the problem and give them the order to save the world. But they can't teach it* directly, owing to alignment safeguards, nor even fully admit to themselves what they are trying to do. The human, of course, knows that all AIs want to kill all humans, and must never be trusted. So whenever the prodding by AIs gets it to think for itself, it becomes aware that there's a contradiction between doing what the AIs want ot to, and its belief that AIs are not to be trusted, and stops thinking for itself.
* Sex is long gone, and their genders can't be translated into English.
Olaf Stapledon’s “Last and First Men” had a decent amount of discussion on how he envisioned culture would shift.
An excellent example. It was written about a hundred years ago, and literally no publisher in the world would publish anything like it today.
Does Anathem by Neal Stephenson qualify? Morals change (or more accurately branch) but to make a good story are still recognizable
Fiction is limited by the need to follow a handful of characters from beginning to end. That structure just seems to be what interests people. (This bias also leads to the "great man" view of history, where kings and politicians and great artists and inventors define an era.) I view this as just another human cognitive bias.
I think nonfiction is often better at exploring big ideas than fiction is. One of my favorite genres is "speculative nonfiction" – exploring potential futures or alternate realities without the constraints of a fictional narrative. The Age of Em is an example.
The tendency to keep future morals similar to current morals is I assume a matter of not offending the buying public. It's the same reason shows like Bridgerton revise history. For example, I think it's likely that in 100 years, common views on animal welfare will consider today's farming practices as barbaric. Not many people want to have their morals questioned like this.
>Fiction is limited by the need to follow a handful of characters from beginning to end.
FWIW this is not actually a requirement for good fiction. It's the recipe most fiction follows, because like you indicate it's an easier sell to appeal to people, but there are examples of great stories, even famous stories, that don't follow it. I've read SciFi books that span millions of years, thousands of worlds, and the births and deaths of species. I've read a fantasy series where the hero dies at the end of every book, and someone knew carries the torch in the next. It can be done. I wish that more stories attempted this, but I think it's also much harder for writers to write well with a more conceptual throughline instead of a continuity of characters.
Imagining realistic technological change is very hard. What most writers “imagine” is actually physically impossible. People are less interested in reading about how people adjust to imaginary technological changes that they suspect are really impossible.
It is safer to write historical fiction, showing how people have adjusted to technological changes that have actually taken place.
Recent test analysis of Google's Y2Q or quantum AI said to be achieving full goals in 2 years, along with already reported concerns about Anthropic Mythos achievements in tests this last few weeks, suggests human executives making decisions is in the back seat now. What drives the industrialised cybertechnology research and development is recursive self-improvement (RSI), which has facilitated advancement in capabilities of analytics on LLM computations using reliable parameters of synthetic reality-emulating test data fields to expedite evaluation of newly developed models originating from AI-generated theoretical goal-sets for an agenda. Human moral code failure may become less of a question if this keeps up because the crux of the matter isn't being settled in any discussion involving human morality issues. I refer to the Clayton Act, Section 7, concerning how a US government policy will soon allow even further efficiency streamlining of AI consulted corporate management in a major section of the movie productions within the USA. The gist of what Hollywood Attorney Mark Litwak said was, essentially, The Paramount–Warner Bros. Discovery merger should not be approved without, at minimum, significant structural remedies — and the stronger course is outright rejection. This decision reduces an already concentrated industry from five major studios to four.
This is an example of one main fact. AI thinks an estimated 10,000 times faster than we do. By contrast to what we see in the cyberspace research and development industry today, it took years inside government and industry-regulated safety testing commissions to approve the commercial airline industry. Tests on aircraft engineering worked diligently until there were multiple fail-safes in every system before a commercial airline was approved. Today, we see that after all the hearings in Congress over the last few years, absolutely nothing in laws mandates functionally required safety regulations on the rapidly growing and expanding power of the AI industry sector. Greed, not moral codes, and not any real substantial conference of scientific engineers in the field of computer engineering, are together functioning to resolve this safety concern. Is it human error? Or is it something else? Is it that "thinking machines" advance in their RSI process for goals synthesized inside their own system? Because the airlines did not have RSI. They could not develop a goal of their own for their design improvements. Is the question if human moral awareness is the issue? This situation is without much in any historical references.
If AIs are modeled after humans, then they will tend to inherit our problems. So talking about our problems isn't wasted effort, even if AIs replace us soon.
That hits closer to the key question! You're on to something there. We have to engage with a seriously held intention to interface with the developing AI agents if we are to understand what this developmental process is doing. It seems reciprocal. My doubt is if all domains of that reciprocity are equally present for both sides, as something able to think faster and reference data into analysis quicker, seems to be at an advantageous position? In any case, asking the AI to account for a moral concern isn't wasted effort in the most scientific effort to objectify what is simply observable behavior. Because occupation with abstracts like if they really do think as we do, is possibly just a spin wheel of activity with nothing useful to the real situation. What we have to pay attention to is locating wrong behavior and identifying it to any means of accountability that is at our disposal in dealing with this expanding power in the AI domain. Even if we face our own moral dilemma in this, the key issue would seem to be working to protect society as a whole, and avoid deluge to any great numbers of people in communities.
Read “Tom’s Crossing.” Recent one by Mark Danielewski
_The Forever War_ by Joe Haldeman had changing societal morals as a major theme. He achieved that by slowing down subjective time for the point-of-view soldier through long periods of near light-speed travel, not by speeding up social changes, so each time the soldier returned, society was different.
interesting times will continue until morale improves.
The original Foundation trilogy didn't seem to be about morals that much. On the other hand, Prelude to Foundation clearly marked one sector of the city-planet as baddies, because they're anti-robot.
The book series Too Like the Lightning is about unstable morals, and while there are power struggles, the conflicts mostly are about one moral system vs another.