Live data from Hacker News

Pre-2022 Books

notes.lorenzogravina.com

31–40 of 141 posts

Re: Pre-2022 Books

#31
C-c C-v existed well before 2022. Most of history until the Renaissances consisted of the bulk of scholars copying out of "the book" whatever the book happened to be (Euclid, The Bible, Aristotle’s Logica Vetus, Cicero's Orations and De Officiis, 四書五經, 史記, 文選)

The liberal concept that the everyman should have their own original thoughts that others should consider is a historically a very new concept. And we start getting things that look a lot like C-c C-v quickly after the Renaissance.

See humans have the tendency to romanticize the past, and if this is allowed to compound they elevate really quite dismal people to the realm of literal godhood in some cases. If you asked someone a thousand years ago what they though life was like thousands of years in the past and what it will be like thousands of years in the future most would have said the past was better in all regards including health, strength, morals even technology; while the future would be viewed as the continual circling of the drain. Put yourself in their shoes, you go look at a Roman Colosseum, you can't build that, nobody you know can build that. If you asked Vitruvius during the construction of the Aqueducts he would tell you that he's maintain the knowledge of his ancestors, whom could have build such structures if they needed them or had the manpower, and the technical problems are just a trifle. If you pushed him, he might invoke Providentia and that if the gods stopped blessing you we'd fall even faster.

This kind of discovered then lost fits better narrative within the human psyche better than the unintuative truth is a constructed social conversation, that can be semi-formal and rigorous (the scientific method) or lax (common sense) depending on the setting.

Re: Pre-2022 Books

#32
post #30

And there might be no way to prove you really wrote something Post-2022. I wrote a long article, all by hand. I never used any LLMs, even for searching. I checked it with a couple of AI detection tools and they confidently said that 60% of the article was written by AI.

The thing is, llm token frequency was derived from human writings like yours, and rlhf for good writing practices, like the emdash. So getting ’detected’ on good writing it’s unfortunately to be expected. My broken ESL english is much safer for now

Interestingly, you're actually more likely to be flagged as AI if English is your second language.

Re: Pre-2022 Books

#33
It’s not just 2022 and earlier books. There’s a supply chain problem as well. I’ve seen two older books so far from Amazon which were AI generated copy text with a genuine looking cover on it. Amazon just took the return and probably restocked it for the next victim.

I tend to buy books from second hand book shops and eBay now and usually older or well used copies. A good sign of their authenticity.

Re: Pre-2022 Books

#34
Hot take: I think it’ll be pretty much the same as it was. If anything it will get better.

You will still have gatekeepers and taste makers. Publishing houses will screen fiction for well-written and interesting fiction. Word-of-mouth, personal recommendations, and endorsements from people you respect will continue to outweigh algorithms, if you care.

For cheap reads, how much of a difference is there between James Patterson’s 734th beach read thriller and what an LLM with a 50m token context window can produce? Does it matter that it’s not written by six ghostwriters? Probably not to the median Hudson News buyer.

For non-fiction, it’s easier to gather research and related materials. If you were cherry-picking facts to make a narrative, yeah, that’s easier, but it’s not like we haven’t gotten really good at that anyway. Again, there will be cooling off periods for scholarship to be debated and coälesce.

What will get better is people asking questions and getting well-researched pieces on a specific niche or confluence of topics. AI is just-good-enough-to-be-dangerous now. It will get better. We’ll learn to harness it (literally) to iteratively fact check and cite sources. We will build repositories with heavily sourced facts for it to build upon. It will be pulling together “truths” that can be traced, then incrementally adding inference across those, which can then be verified and are a new fact.

I read a lot. I love, love, love new and original authorship. I deeply value writing as a craft. There will be a lot of garbage. More than there is now, at an incredible rate.

And we’ll figure it out.

Re: Pre-2022 Books

#35

And there might be no way to prove you really wrote something Post-2022. I wrote a long article, all by hand. I never used any LLMs, even for searching. I checked it with a couple of AI detection tools and they confidently said that 60% of the article was written by AI.

I track every change in the fiction I write in git. Its not hard proof but it at least shows how the prose evolved over time and is something like a proof of work.

Re: Pre-2022 Books

#36
You are not alone. I like to read Harry Potter fan fiction [1] and I have started checking the publication date when Im searching for something new to read. I started doing this passively and realized it after the fact.

Have you ever met someone who could say all and do the right things but never made you feel anything, or your gut was sensing an ulterior motive? It's a magic trick we are all bewitched by at some point in our lives. I suppose I filter by published year because I dont want think about if I am being tricked or not.

[1] There are some very talented writers[A] out there who (I assume) cannot do the world building part.

[A] Recent Favorite: https://archiveofourown.org/works/1134255/chapters/2292768

Re: Pre-2022 Books

#37

I haven't seen any minimal sign that any of the fiction books I read lately was LLM helped. Writers seem like a particularly anti AI crowd too. Has anyone? Now I'm curious if it's just my particular bubble.

What makes you think you'd recognize it? Do you work a lot with LLMs for fiction writing?

More than you can imagine. OKRs, brag documents, quarter reviews....

Kidding aside, I would be surprised if something larger than using it as a thesaurus/corrector is slipping by. Literature is genuinely hard.

Re: Pre-2022 Books

#38

And there might be no way to prove you really wrote something Post-2022. I wrote a long article, all by hand. I never used any LLMs, even for searching. I checked it with a couple of AI detection tools and they confidently said that 60% of the article was written by AI.

Most "AI detection" tools are BS, report AI writing for all writing. The current issue with AI writing is that it has a very generic, easy-to-spot style if you spend even a bit of time working with it. If everyone in the world spoke in the same manner, used the same punctuation, spoke in exactly the same catchphrases, the world would lose its richness, and that is the problem with AI writing and communications in general - it has 0 personality, and humans by nature engage with strong and unique personalities. Over time, people will realize the futility of using AI in creative writing, or AI will get really, really good at not just sounding human but being human.

Yet, from what I can see, AI writing is mostly used by people who don't know a thing about writing, and because they have bad taste, they do not see what's wrong with AI writing and put it out there.

At the end, you write for a purpose: for marketing copy, etc., you would require a different type of writing talent than something like writing a fiction book. But AI doesn't understand this nuance; it has only a default type of communication, which is highly optimized for being a chatbot. It is possible to write a very good text using AI if you have taste and you know what you're doing, but most people don't.

Similarly, a lot of vibe-coded apps are garbage, but because the people creating them lack software domain knowledge and don't even know what they don't know, they think it's good and put it out.

We have a massive problem here that's not just limited to writing - the promise of AI for the mainstream market is that you can replace domain-specific knowledge and have world-class execution in any vertical with just AI, but that's very overhyped imo and doesn't stop the people who don't have domain experience to try out stuff with AI and not realize what they made is a steaming pile of shit in reality.

Re: Pre-2022 Books

#39
post #34

Hot take: I think it’ll be pretty much the same as it was. If anything it will get better. You will still have gatekeepers and taste makers. Publishing houses will screen fiction for well-written and interesting fiction. Word-of-mouth, personal recommendations, and endorsements from people you respect will continue to outweigh algorithms, if you care. For cheap reads, how much of a difference is there between James P…

My worry is less about scholarship than the next generation of readers and authors. It is too easy to be lazy right now. Too easy to skip the difficult work of struggling with ideas. Yapping with Claude probably (?) doesn’t have the same rate of retention and reinforced learning _in humans_ as digging through source material and writing by hand.

Growing critical thought, in my experience, has always been the much harder problem. Not sure we’re in for a good time on that front.

Re: Pre-2022 Books

#40
post #30

Earlier quoted context omitted.

The thing is, llm token frequency was derived from human writings like yours, and rlhf for good writing practices, like the emdash. So getting ’detected’ on good writing it’s unfortunately to be expected. My broken ESL english is much safer for now

Interestingly, you're actually more likely to be flagged as AI if English is your second language.

That tracks with reality, as the majority of people don't have English as their first language. Depending on data sources used for training that could well reflect into AI detection tools.
Post reply on HN