Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Warranty Void If Regenerated

nearzero.software

Post reply on HN
Print view View original

301–310 of 353 posts

Re: Warranty Void If Regenerated

#301

Post by stingraycharles »

ffsm8 wrote: post #276
stingraycharles wrote: post #240

Earlier quoted context omitted.

Because you need to do some pre-filtering on where to focus your attention, and you want to make sure the author put some thought into the article without having to analyze it. Due to LLMs making the cost of publishing “thoughts” extremely low, there’s now an over-supply of content that looks decent on the surface, but in reality the author has probably spent less time on than the reader.

Are we ready so far down I to the LLM denial mindset that we consider an author spending multiple months crafting this to be "worthless" and less investment then your casual reading?

No, I believe this is a great post. It’s awesome. Even more so because it’s AI generated, as it shows what AI can do when given a lot of quality material to work with.

I’m just talking about the general topic about the usefulness of an “this is AI generated” classifier.

Re: Warranty Void If Regenerated

#302

Post by ijk »

intended wrote: post #247
Aditya_Garg wrote: post #226

Earlier quoted context omitted.

I’ve said this many times before AI is just a tool If you used a fancy auto bake cake machine instead of an oven, you still get to claim that you made the cake. 100 years ago someone would be making the claim that using an oven to make cakes “doesn’t count” All AI did was raise the bar It’s quite clear here that the author spent a lot of time on this so he absolutely gets credit as the author

Eh. Why would I give him the same credit I would give a writer. Or why would I give a writer the same credit I would give someone who created the AI prompts and scaffolding to generate this? Being unhappy about not being able to call oneself an author, ends up betraying a lack of confidence in the work or process. In the end writer, dancer, actor, whatever - these titles come from their impact. There will be a differ…

Ironically, the story can be read as gesturing in that direction, as it's ostensibly about giving a new title to a particular job.

In general, though, I think part of the mistake people keep making is that they try to imitate what would be value to engage with if a human wrote it, in an attempt to claim the role of an author of a book or whatever. There's likely artforms that are unique to what an LLM can facilitate, but trying to imitate human artforms is going to give you stunted results. The AI is very good at imitating the form but not the substance.

Once we stop trying to generate and pass off AI essays, novels, choose your own adventure stories, and all the other human genres as being human writing, we'll have a chance to figure out actually interesting artistic forms.

Re: Warranty Void If Regenerated

#303

Post by ijk »

KronisLV wrote: post #219
toofy wrote: post #208

Earlier quoted context omitted.

because we typically want to know the writer of a piece. we want to know where to lay credit. every book you buy has an author credited. articles in newspapers and magazines have photographer and author attributions. asking an ai to write you a story does not make you an author. if you ask someone to take a photo for you, you don’t magically get to say “look at this photograph, i’m a photographer.” if you ask someone…

> because we typically want to know the writer of a piece. we want to know where to lay credit. Does the average person really do care all the time? Maybe the outlet it comes from as a whole (factuality, political lean) but more rarely the exact author. Many don’t even have the critical skills for any of it and consume whatever content is chosen for them by whatever algorithm is there. We probably should care, I just…

In Peter Watt's Blindsight, the aliens understand language as spam, a hostile intent to waste their time, and respond by opening fire.

Reading LLM slop without warning makes me see their point of view.

I think there's useful ways to engage with LLM writing, but they are often very different than human writing.

A human writer, a good one, often has ideas that are denser than the words on the page, and close reading is rewarded by helping you unpack the many implications.

With AI writing, there's usually fewer ideas than words, and so it requires a different kind of engagement. Either the human prompter behind it didn't supply enough ideas, or they were noncommittal enough that their very indecision got baked in.

LLMs are very prone to hedging and circling around a point while not saying much of anything. Maybe it is the easiest way to respond to RLHF incentives and corporate-speak training data. Or maybe they're just intrinsically stuck on being unable to find the right next token so they just endlessly spiral around via all of the wrong ones. Either way, there's often a whole lot of cotton candy text that dissolves when you try to look at it more closely.

Re: Warranty Void If Regenerated

#304

Post by ijk »

throwaway2037 wrote: post #289
NateEag wrote: post #265

Earlier quoted context omitted.

Well-put. This "flavorlessness" is all over the story, and paired with the obviously genAI images is how I realized as I read that this was either generated or at the least deeply driven by AI. It constantly described facial expressions, tones of voice, and other emotional cues in generic, dry terms that communicated nothing but the abstract notion of "this person felt a particular way about what happened and it's up…

> obviously genAI images Five years ago and before, the blog post author would have gone to Fiverr and asked for an artist from a developing country to create some illustrations. There are many, many images on the Internet from five years (and before) that look similar. I object to your use of the adverb "obviously".

No, I clocked the AI images before I noticed the text. I think the "obviously" is earned.

You are correct that a previous era would have included a bunch of Fiverr images that would be in sort of that style, but it's not the style that's the problem. None of the images say more than the text that they're illustrating. It's subtle, but once you notice the lack of information density it becomes starkly apparent.

Re: Warranty Void If Regenerated

#305

Post by CamperBob2 »

sebzim4500 wrote: post #275
taneq wrote: post #186

Earlier quoted context omitted.

I think two sane things. 1) It’s good in the long run that they didn’t prevail at that time. 2) They did actually, in fact, have a point .

I mean obviously they had a point? No one wants to lose their job.

Everybody wants to lose their jobs. Almost by definition your job is something you do not because you want to, but because you need to earn a living. Even if your job coincides with your hobby, you would prefer not to have your economic welfare tied to it in a way that drives how you engage with it.

We are on the verge of making this possible, if a bunch of myopic morons -- people who have never been right about a single long-term trend in history -- can be convinced not to screw it up.

Re: Warranty Void If Regenerated

#306

Post by ijk »

mwigdahl wrote: post #291
fzeroracer wrote: post #207

Earlier quoted context omitted.

Because 'quality' is a misnomer. LLM writing has quality in the same way that a press release from a big company has quality, or a professional contract written by a lawyer has quality. It is functional, generally typo-free and conforms to most standards but that doesn't mean it has flavor or spice to it. Creative writing is the intent to convey feelings, thoughts, to create atmosphere. Here's a great example of the…

I took that phrase differently. The story makes the point that the AIs fail when metrics of quality can't be expressed in words. The use of a bare "adequate" reinforces the opacity of the coffee's quality. Certainly it would have worked well to use more words to convey specifics of the "adequacy" as you mention, but IMO that would have undercut the link back to the theme of human ineffability. Obviously everyone's mi…

Adequate coffee almost works as an image.

In the hands of Douglas Adams or Kurt Vonnegut it could be spun into a whole recurring motif.

In this case it's merely...adequate. Almost captures the density of ideas packed into something like "The ships hung in the sky in much the same way that bricks don't" but doesn't quite manage the same effect.

Re: Warranty Void If Regenerated

#307

Post by ijk »

fluoridation wrote: post #152
rustystump wrote: post #149

Earlier quoted context omitted.

No, code is a way of communicating ideas, or more correctly information. All languages convey information. All languages convey ideas.

Did you read past the first sentence? The kind of information that a piece of code transmits is fundamentally different from that which is transmitted by a sentence or a song.

Yes, though I would take it in a different direction and say that LLMs are better at putting actual ideas into code. They've never gotten real feedback on how their literary metaphor feels, but they have gotten very direct feedback on whether code runs at all, and slightly more indirect feeds on whether it runs as part of the larger system.

Re: Warranty Void If Regenerated

#308

Post by ijk »

BoorishBears wrote: post #97
donatj wrote: post #61

I'm trying to sort out my own emotions on this. I did not realize this was AI generated while reading it until I came to the comments here... And I feel genuinely had? Like "oh wow, you got me"... I don't like this feeling. It's certainly the longest thing (I know about) I've taken the time to read that was AI generated. The writing struck me as genuinely good, like something out of The New Yorker. I found the story…

I suspect (but don't know) that this had to be edited somewhat heavily or generated in isolated chunks: I've generated a lot of fiction with Claude and it has a chronic issue of overusing any literary device one might associate with good writing once it appears in the context window I think if you left it to its own devices, some of the narrative exposition stuff that humanized it would go off the rails

Yeah, there's often a heavy instruction and recency bias that just squeezes all of the nuance and subtlety out if it.

Re: Warranty Void If Regenerated

#309

Post by alexjplant »

donatj wrote: post #61

I'm trying to sort out my own emotions on this. I did not realize this was AI generated while reading it until I came to the comments here... And I feel genuinely had? Like "oh wow, you got me"... I don't like this feeling. It's certainly the longest thing (I know about) I've taken the time to read that was AI generated. The writing struck me as genuinely good, like something out of The New Yorker. I found the story…

My $.02 is that in the domain of software engineering LLMs have largely automated the process of copy-pasting from StackOverflow and existing parts of the codebase. Architecture and product management is still very necessary. In the same fashion they can also automate writing a novel. The issue is that prose is sometimes much more important in literature than it is software (because, after all, users use software, they don't read the code). I say "sometimes" because this clearly doesn't apply to stuff like schlocky bestsellers that one buys in airport stores and reads like movies.

When ChatGPT first came onto the scene I actually started using it to write something in this vein - a techno-thriller starring a former fashion model trained in Krav Maga working as a nuclear physicist who discovers a sinister government conspiracy to alter the foundations of quantum mechanics and enslave humanity with assistance from extraterrestrials. And, of course, only she can stop them with the help of a gruff-but-sensitive retired Marine who has since opened a ranch where he teaches orphaned puppies calculus. I only got 20 pages (so one gunfight and a car chase) in but it was as riveting as anything. Context limit cut my efforts short. Perhaps I'll revisit it soon.

I say all this to say that if words themselves are distantly secondary to narrative then I don't see anything particularly wrong with leveraging an LLM to help crank something out.

Re: Warranty Void If Regenerated

#310

Post by PaulHoule »

'the concept of “broken software” had been replaced by the concept of “an inadequate specification,”' represents a fundamental misunderstanding which has been a source of trouble in the industry for a long time.

That is, a lot of "broken software" has always been rooted in "an inadequate/incorrect specification" If problems in the spec are discovered up front they are cheap to fix, the further along you go in development or deployment, the more expensive they are to fix. AI doesn't change that. Like maybe with AI it is 20% faster to fix [1] across the board but it is still more expensive to fix things late -- you might think you are done with waterfall but waterfall is not done with you!

[1] My 20% is pessimistic but if you think you are 10x as productive with AI at putting functionality in front of customers in the long term with a universal scope I believe you've got the same misunderstanding about product life cycle that I'm talking about

Post reply on HN