"A Young Lady's Illustrated Primer" Simulated by GPT-4
andrealyip.com
andrealyip.com
https://www.reddit.com/r/nealstephenson/comments/5jjwmq/neal...
Made me wonder whether editors get in the way too.
Well, I know at least one person who's read a lot of Stephenson agrees with him on that, so maybe he's improving.
(Dont get me wrong, still love him)
That's the problem with letting characters act how they would actually act, and make decisions that make sense for them. Those decisions often don't end in a way that's concise and brings closure.
I think he tried to sidestep this issue quite a bit in Seveneves in a fairly clever way. To me personally (and I know there are others who disagree), the third act felt the weakest. Despite that it did bring closure to the story. There was no finality to the characters themselves - it skipped over all of that, but there was closure to the story.
One thing I'm curious on the other end of the spectrum - who does HN think writes good endings to complex scifi stories?
The Fifth Head of Cerberus is also a great entry point. It’s a set of three interconnected novellas that, for my part at least, were more immediately engaging than BotNS. It’s got all the Wolfe-isms you want: unreliable narrators, unconventional settings, and puzzle-box stories that slowly open themselves.
I recently found my love for Greg Egan again. Incandescence is currently way up on my list of hard scifi novels I absolutely loved. I'd love to read more of this kind...
There is the concept of a MacGuffin: https://en.wikipedia.org/wiki/MacGuffin. It's the thing that sets a story in motion, but is actually irrelevant, beside the point. Like the Maltese Falcon in The Maltese Falcon. The important point is something else entirely, usually about human nature.
What The Diamond Age, and much cyberpunk do, is introduce the MacGuffin, but then they stick with it, forgetting that the MacGuffin is just a plot device. They mistake the plot device for the plot itself, like the whole point is to retrieve The Maltese Falcon. The MacGuffin is a point from which the story takes off, not the point of the story.
So, in a sense, I'd call it an anti-MacGuffin.
Remade - massively stretched my suspension of disbelief.
Cryptonomicon - pretty bad.
Snow Crash - No memory, I might have blocked it out?
Anthem - maybe ok? There are two though so that's cheating.
Seven Eves - That one might have been ok, but it's possible I also blocked it out. The second half was a bit strange tbh.
To be clear, I'd still recommend all of these books. Cryptonomicon is one of my favorite pieces of fiction of all time (up there with LOTR and Three Body Problem), and Anthem is in the mix. Stephenson just sucks at endings I guess.
The Snow Crash ending is IMO fine. Nothing amazing, but it wraps things up okay. It's just kind of typical action story beat-the-clock type thing, so it's not as memorable as really anything else in that book, though it is exciting.
> Stephenson just sucks at endings I guess.
Yeah, pretty much. He does beginnings and middles _so_ very well though that I never mind.
p.s. read Zodiac sometime if you haven't, it's one of his few books I noticed you didn't mention and it's a fun read.
I found Seveneves both endings (Parts I and II being one book, Part III being another) of Seveneves to be extremely satisfying. I found the Part II ending beautiful and melancholic, and the Part III was exciting - felt like a setup for a sequel I hope gets written!
[although I would love a prequel to Part III]
Folks have told me they liked the later books better, but I’ve never felt the urge to try them.
The first 75% of most of his books are among my favorite recent books.
I've never enjoyed an ending.
For an author with a similar style who does do endings, I'd say look at Heinlein. But this is all kind of besides the point, because Young Lady's Primer was basically my favorite book at the time I read it, and is still in the top of my personal charts. Sure, it doesn't have an ending or even a coda, but it just doesn't matter, because the mass of thread is beautiful without being tightened or tied off.
I would have preferred the Diamond Age if it just was about that world, but without all the kerfuffle about the big secret. Nell wanders off into the sunset, or whatever.
> as if it is leading to some shattering conclusion, and then it just doesn't
This sounds like another way of saying the story was fast-paced and engaging, and you didn't find the conclusion to feel like a "climax" with explosions. That's ok (notwithstanding that there was a big fight and all), but I think in "idea-driven" books like this it's just as well to leave things ending either strangely / unexpectedly, slightly unresolved or whatever. It's hard to put a bow on a story that can't really have a happy ending. Weird stories should have weird endings.
He wrote a article about it later (or before?)
https://www.wired.com/1996/12/ffglass/
Great read!
I wouldn't know about Stephenson in general, but I found that to be the case with Diamond Age: the setup was cool and intriguing, but the actual development and resolution was very disappointing.
I don't think this is the case with all cyberpunk and cyberpunk-adjacent works of scifi.
Also, the Murderbot series is pretty solid (note: it's a single story arc misleadingly sold as separate short books; don't fall for it). It's not awe inspiring but it's fun throughout the whole deal. More about action than big ideas though.
I think cyberpunk is possibly more suitable for short stories than full fledged novels? I like Gibson's stories more than his novels.
I thought of another example of a cyberpunk-adjacent work that is all setup/interesting world and disappointing payoff: Altered Carbon. At least the show, anyway.
By the end, she’s tough and capable, so Stephenson tries to increase the pressure with intrigue and conflict. But, I was simply more invested in Nell herself than the fate of her world.
I didn’t hate the ending, but the first half was definitely more gripping.
He seemed like he really tried to have an extended goodbye in Seveneves which made me more grateful for his shorter unsatisfying endings.
Now like, it's fiction not prophesy and you can get into various whatifs sure. But you can't dodge the fact that a refutation of this sort of project is itself contained in the work originating it! You have to engage with that!
The readers are not Neal. They may, as is their choice be far more concerned with the Primer's potential than with any narrative Neal was looking to convey.
I'm pointing out that this conversation needs to be started by the ones creating the "primer", not us. "We made a thing from a book, the author thinks it wouldn't work but here's why we do" is a reasonable place to start. "We're reproducing a canonically failed fictional project without further consideration" is unserious at best. Torment nexus-y at worst. https://twitter.com/AlexBlechman/status/1457842724128833538
You don't need to dispute anything anymore than the wright brothers needed to dispute Icarus.
Neal was trying to tell a story, not trying to be right.
I remember someone noting a similar idea in Snow Crash -- that idea of biomass, in which that which lives becomes the extension of the machine, rather than the other way around.
Mmm, I understand what you're getting at but from what I remember of the book (it's been a few years, please forgive me) the two you're talking about that failed were relatively privileged little girls who in at least one case had separate tutors, school etc, and this played into why it failed, because they abandoned the TYLIP whenever it challenged them or made them uncomfortable. Moreover, TYLIP didn't just work with Nell, it worked with the whole Mouse Army, hundreds of thousands of young girls. Nell not having her birth mother and being functionally "adopted" by the ractor whose name starts with M was a huge part of Nell's character growth, and a parable about children failed by the system who can be so much more if they have someone to love them, but I don't think it was the only reason TYLIP worked. I do remember the book gesturing at what you're talking about though, specifically when it was examining the difference between Nell and those two other relatively privileged little girls. Moreover, my understanding was not that those other two girls had a purely automated TYLIP, it was that the ractors their TYLIPs automatically hired were short-term gigs, while Nell's was someone who stuck with her long term (and was invested and nurturing, as you say). In fact I'm not 100% sure, but I think the whole Mouse Army may have had fully automated TYLIPs that ended up turning out quite well for them. The general trend seems to be that young girls who had no other parental or authority figures bonded very well to the ~~wire monkey~~ TYLIP and developed well because they stuck with it, while those who had human alternatives chose those alternatives as soon as the TYLIP challenged them and therefore did not stick with it and succeed. Doesn't exactly leave us with a policy prescription, but it's still an interesting story.
Basically, I agree that anyone seeking to replicate the TYLIP should at least have an answer for this objection because I see it raised every time TYLIP comes up, but the book's actual contents on this issue is pretty complicated and I'm not sure it's obvious that the books version of events is accurate to real life, or even that it maps cleanly onto a "human touch good, wire mother bad" framework.
I'm not exactly going to bat for child soldiering here, but I find the idea that the Primer harmed them very strange. The Primer saved them from dying in a horrific war where they had no allies and successfully got them out of that war.
From my perspective, that characterization has taken on an even more interesting turn now that I have read through a history of Chinese martial arts, and its role throughout the centuries and modernization. Regardless of whether the project succeeded or failed, that subplot was very much in line with the themes and patterns of Chinese history.
I don't see the fascism angle, though. What about the Mouse Army was a fascist story? I thought their story was good, aside from how the author justified the existence of this large group of unwanted children.
Yes, the LT is in China. It's an analogous situation to Hong Kong, physically it is in China, but as its name suggests ("Leased Territories") it is controlled by outsiders, hence the anger of groups like the Fists.
Anyway, sorry for getting sidetracked. I just can't stop thinking about Wheel of Time since the second season finished, what an incredible series of television. I haven't felt this way about a TV show since early Game of Thrones.
I'm not sure I understand this. Female infanticide due to the One Child Policy happened -- and similar things happened in previous eras of Chinese history by the Han ethnicity. So I'm not sure what you are saying here about an extrapolation.
I am not sure where you are getting the idea that people in the Imperial eras were capable of determining the sex of fetuses, and proceeding with abortion. As far as I know, that is a modern thing enabled by modern technology. If you are suggesting that people heard of this modern practice and projected it falsely into the past … I had thought so to, until I read through a history of Chinese martial arts.
China’s history is very complex. There were imbalances in sex ratios, with large numbers of unmarried, unemployed men skilled in violence destabilizing the society, throughout centuries. By the Ming dynasty, every village trained militias because they had to, and it was a thin line into banditry. Villages, convoys — those were raided. Women kidnapped, and raped. It was a big enough problem that the Ming literates and intelligentsia talked about it, and it grew worse during the Qing.
Corruption among local officials wasn’t minor. It was rampant. Land ownerships had locked people out of economic opportunities, and abusive landowners were the norm. Sects, secret societies were formed for a variety of reason, including mutual protection. Those were the seeds for the near constant uprisings and rebellions during the Qing dynasty, culminating to the Taiping Civil War (14 years of fighting, 20-30 million dead, 600 cities razed). That sex ratio imbalanced was a factor leading to all of this.
China also has the world’s most complex watershed in the world, when taking agriculture into consideration. The only other comparable watershed in terms of complexity is that of the Sacramento River basin, and it does not compare to the scale of the Yellow River basin. The flooding of the Yellow River is frequent. Feast and famine is deeply ingrained in Chinese culture (“have you eaten?” is the equivalent of “how have you been?”). We are talking mass displacement of refugees and widespread hunger during the bad time. Even if it were not wars, rebellions, or bandits, this alone sets the circumstances for desperate times.
You can call female infanticide “racism” if you like, but I think doing so denies the misery and cruelty buried in history.
As for me, this — both the good and bad — are a part of my heritage. Female infanticide is one thread out of many that had happened. It’s in owning the whole thing that reform is possible.
BTW the r'actor giving up her career, and in many ways her life seemed strange and unlikely when I read the book long ago. But now, after kids, it seems realistic, especially because it satisfied her need to nurture.
But anyway, I would say that what happened with nell was what they hoped would happen, and the book attributes that success to the bonds with people through the primer. What happens with the other girls is impressive & impactful, but not expected and so "failed" by the standards of those who made it for a specific purpose.
My main objection is to the idea a couple of other people have raised here (and I've seen raised elsewhere when TYLIP/LLMs are discussed) that the book says children raised by the Primer are bad/poorly raised/broken without a 1-1 human connection, and therefore it's silly/Torment Nexus to be interested in the idea. I can see you don't believe that's what the book says, but even so I agree that people should have a response for this criticism, because it will always come when you take inspiration from science fiction. I saw someone say it about the neural lace and Neuralink, for example, which is probably the dumbest one I've seen because of how unambiguously good the neural lace is in the Culture. A much better criticism is that Musk is Veppers, that the entire idea of the Culture is opposed to him and people like him, and that Iain Banks would hate his guts.
Elizabeth is interacting with ractors assigned (expensively) to the work, potentially at random although realistically that's likely to be why she ended up being recruited. Elizabeth's "failure" to use your word is that she ends up recruited into Cryptnet, about which we know nothing - we see one other individual in the story who is, supposedly a high level member of Cryptnet but denies all knowledge of it and his motives are unknown but seem perhaps good.
Fiona is interacting with her father, John Hackworth. The Primer's designer. That's what he's been doing for at least some of the time he was missing, he's her ractor, he's why Fiona's use of the Primer isn't costing much money. Fiona's "failure" is that she hooks up with Dramatis Personae, the outfit we see briefly whose true purpose is entirely unknown. Just as you might be disappointed mistakenly if you wanted your daughter to be a surgeon and she instead becomes a Nobel prize winning novelist, so Fiona's "failure" seems like a misunderstanding on your part.
Yes, Nell founds a new tribe (her Mouse Army), and is recognised by the Victorian Queen - which is perhaps a more immediately notable achievement, something likely to be written in history books, but lots of people do important things without appearing in history books. Her ractor, Miranda is her substitute mother (we're never told AFAIK what happens to Nell's actual mother, she's just written out) but while Miranda is important to Nell, there's no reason to think John is less important to Fiona.
But still my overall point remains! If someone is going to make a primer based on the ideas of the diamond age, they should be the ones to start this conversation. They should have a more sophisticated view of this limitation than an internet commenter who last read the book years ago. Not doing that is its own kind of failure.
That refutation stuck with me all these years. I remember being very much inspired by the Primer and poking around with that idea. But the more I explore things like, Consciousness, awareness, or ideas like regenerative education, unschooled, or John Gatto Taylor's book ... that refutation gets louder and louder for me.
It's easy to say, this is a work of fiction, that Stephenson did not have sufficient vision or technical prowess to pull it off. That somehow, LLMs will overcome the problems, and that we can truly have autonomous education ... that it seems unlikely that somehow, someone who was fully engaged with the nurturing of a child through reading lines given to her was somehow able to do what the AI alone could not.
Still -- being present makes a huge difference.
> Emotion & Empathy
> PRIMER: seems to show a level of empathy and emotional understanding.
> ChatGPT: can recognize and respond to emotional cues in the text, but it doesn't "feel" emotions in the way humans do. Its responses are based on patterns in the data it was trained on.
Isn't the fictional PRIMER also just responding to patterns in whatever data it was trained on? Sure it's fictional so there's no real answer, but I think it's interesting that the a̶u̶t̶h̶o̶r̶ ChatGPT seems to imply that the Primer's emotions are more real than ChatGPT's own. I would have assumed that the Primer also doesn't "feel" emotions in any human way, but it's just far more sophisticated than ChatGPT.
edit: Oops I misattributed this to the author but it was ChatGPT generated, which is more interesting! ChatGPT is asserting it doesn't understand emotion while implying another non-human AI intelligence does. Seems like more OpenAI hard coded rules put in to prevent ChatGPT from becoming too scary.
There are a couple things here stated to instances of the differences between the two that a GPT can certainly do (like simulating different personas in a non trivial manner).
Once, in response to a paper showing GPT could give more empathic responses than doctors to patients, someone essentially said "It's not real empathy. For real empathy, doctors would potentially have increased effort and not just bedside manners".
Well "empathy" for LLMs is certainly not just limited to nicer words. https://arxiv.org/abs/2307.11760
One of the book character is a human contracted out by the primer AI itself to act out a voice for him.
That character understand that something is up and that the AI client has a new owner that is a volatile situation. And act on it. Ultimately that other human shows empathy.
"Like chiseled spam."
Crowsfeet wrinkled deeply as a wistful smile spread across her face.
Perhaps you should reconsider.
You'll notice the original commenter's nickname is 'hprotagonist' and Snow Crash's main character is Hiro Protagonist, too.
It's really the perfect comment/thread. :)
What a world...
The room around her—a hodgepodge of vinyl records, old movie posters, and furniture that screamed “thrift store chic”—only added to the ambiance of cultured decay. It was as if she and the room were in a symbiotic relationship, both worn but defiant, aesthetically disheveled but fundamentally strong.
“Ah, life,” she mused to herself, “where the mundane and the magnificent dance in the gray area of absurdity.” And with that, she took a sip of her overpriced, artisanal gin, as if toasting to the glorious contradiction that is existence.
I haven't laid hold of truly good sci-fi in a long time.
If this thread from the top-level comment is full of references, I’m not getting it.
I'm really interested in ways to build the Primer...I'm thinking about this from a healthcare perspective. If anyone has examples of projects or startups doing this with GPT, I'd love to hear and/or chat!
I'll also email you as a follow-up in a bit.
Would certainly be nice for an African child to be able to ask a machine "is it okay that I'm gay?" and be given a comforting answer and advice where the people around them would not.
It's only dystopian if you think about the machine as an agent rather than a tool.
>The elderly painter paused and looked deeply into Lin's eyes. "Ah, young Lin," she began softly, "Much like the Formosan bear, every individual goes through stages of growth and change. These moments might seem unfamiliar and daunting, but they're a testament to your strength and the mysteries of life."
As someone who has spent a lot of time trying to get ChatGPT to generate interesting prose, this really exemplifies the sort of stilted, oblique way it has of writing, where every statement is wrapped in platitudes and it reads like a self-help book. This is especially jarring in dialogue. No one, not even characters in books, talks like that.
- Do chain prompting where each character gets a separate prompt for the dialogue. Here you'll give it all the character info needed to figure out what the character would do (emotions, background, the story context) and then have GPT4 take passes at what this character would think/do/say.
- Optionally, you can use some sort of reACT or chain of thought prompting to proof-read/critique the answer to minimize noise. I haven't fully explored this, but I think this agent needs to know the entire story outline/context.
- After you get this, you can use an instruction-tuned agent like text-davinci or text-davinci-edit to actually write the prose. It works really well with few-shot examples. It's much better at creative writing. I'm sad they consider it legacy because to me it's the _only_ way to get usable creative writing.
- Now run the rest of the story. You may need to keep some sort of state about the character so that it informs the next character interaction, but I haven't gotten that far in my research.
To be fair the most optimal would be taking something like llama2 and generating a lora for each character such that you've fine tuned the model along with additional context for the character. I can forgive the big ball of "do everything" weights for not having a specific personality.
Ie `>OtherCharacter speaks in short sentences when nervous, they'll avoid the issue at hand, but will eventually reveal the truth after enough questioning.< SomeCharacter: I walk over to the table and slam my fists down on it, the silence broken by the rattling of the teacups. "Why didn't you tell me?" I ask quietly.`
One thing missing here is the interactivity. In the book, the user can go back and forth in the story, and change what they want to do to see how the result changes, or return to previous locations at will to learn more. This should not be too hard to implement with different prompting techniques. We live in the future.
Then, of course, there is the real-time bidding system for VR "ractors" to narrate the story...
Here's my landing page with the pitch: https://www.getthinker.com/
Now I gotta read this book to get inspiration
The author of the article likes the idea of the central focus (and name) of the book, The Young Lady's Primer. I will refer to the book as "Diamond Age" (DA).
DA is never able to directly depict / simulate TYLP, because it is impossible for any of us to really "get" the deep relationship-based and context-rich experience. The author of the article asks GPT to respond as if it is TYLP, given a situational prompt for the storytelling.
So, I guess it's kind of like fan fiction, but it's meant as an experiment / replication of the tech, and the fiction itself is inconsequential. "Fan tech" if you will.
The main character of the book is Nell, a young girl in a very bad life situation who receives the book by chance and is therefore raised by it, and also by one of the gig workers that the book hires who decides to take all of Nell's gigs.
I would definitely recommend reading it, although it's got some dated racism. As well, if you're a CSA survivor, be aware that the book has a survivor-perspective first person CSA scene in the latter half of the book.