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

The 100k whys of AI

lcamtuf.substack.com

Post reply on HN
Print view View original

41–50 of 111 posts

Re: The 100k whys of AI

#41

Post by SXX »

thinkingemote wrote: post #19

On HN many comments under many threads are about whether the submission was written by AI. You could say I have noticed a pattern in Hacker News comments! In these comments there's a common pattern where some users argue that they do not agree that the submission was LLM written and they often focus on specific details to refute it (e.g em-dashes) and some users see the overall pattern clearly that it's totally obvio…

I'm not worried abou LLM written content, my problem is not word prediction. My problem with it pretty much like with mass produced self help books decade ago.

Good human writing especially on highly technical topics its usually compression of information.

Like you have some experience you want to share with others and you work your brains try to put it into concise story.|

Problem us: AI generated texts are opposite 99% of the time: author usually have bullet point list to feed into machine to add hallucinated word predicted story on top of it.

So signal to noise ratio is much worse.

So reading AI texts is pretty much like listening for stories from humans with mental problems - no one really wants to listen to hallutinations even if somewhere inside there is some useful information.

Re: The 100k whys of AI

#42

Post by Lerc »

I think for that instance to carry weight you would have to provide evidence that the mosaic of books were the product of different people using AI. If it is just one person doing variations on the same thing then it wouldn't mean very much.

Re: The 100k whys of AI

#43

Post by zarzavat »

roncesvalles wrote: post #36
ekianjo wrote: post #34

Earlier quoted context omitted.

prompts will give very different results. this is where you do the work.

Yes but not very different results (unless you're adding new information to your prompt or reducing some ambiguity). Prompt engineering is mostly pseudoscience.

What we need is steering so that we can have models with different personalities, not just different prompts (because context is subject to forgetting), but this will never happen with closed-weight models, I'm not sure if it's even feasible at scale.

Yet another reason why the future is open weight.

Re: The 100k whys of AI

#44

Post by bigbangcmbr »

smusamashah wrote: post #27
dlenski wrote: post #2

A nice illustration of the homogeneity of LLM responses. Another way to describe this effect would be… If you ask humans to write 1,000 books, you're asking 1,000 different humans with different experiences and different skills and different moods (etc.) to write those books. But if you ask LLMs to write 1,000 books, you're probably only talking to 3 or 5 different models, tops. And they've all trained on the same or…

Reminds of Pluribus.

Pluribus is kinda different. An LLM cannot wander too far from the average. Even if it wanted too. In pluribus, the 'others' work toward a common goal, each utilizing their own expertise, knowledge and experiences in a shared way to achieve a common goal. Each is unique. They can, if they want, perform as the host's individual before the the joining. To put it other way, the other in pluribus are convergent by choice, llms are convergent by design.

Re: The 100k whys of AI

#45

Post by NitpickLawyer »

dlenski wrote: post #2

A nice illustration of the homogeneity of LLM responses. Another way to describe this effect would be… If you ask humans to write 1,000 books, you're asking 1,000 different humans with different experiences and different skills and different moods (etc.) to write those books. But if you ask LLMs to write 1,000 books, you're probably only talking to 3 or 5 different models, tops. And they've all trained on the same or…

> A nice illustration of the homogeneity of LLM responses. [...] And they've all trained on the same or similar data, and are trained to respond in very similar ways.

I mostly agree, but this is a very simplified explanation. The models are indeed trained to respond in similar ways, for "basic" prompts. And that's as much a feature as it is a bug. In other words, the bug becomes apparent only if you give 100+ basic prompts. But giving it 100+ basic prompts and expecting originality is a silly endeavour. That's not how you get originality.

The way I'd go about to generate 1000 books, while expecting different outcomes is something along these lines (and nowadays you can ask your favorite LLM to wire up this workflow for you, with decent outcomes):

1. Ask for a list of 20 features that define a book (genre, style, number of characters, tropes, plot, continuity, relationships, etc.)

2. For each feature, ask for a list of 50 examples, ordered from most common to the most unique.

3. Randomly pick 10 features, and for each pick one of the 50 generated items. Ask for the rest of the features to match the theme.

4. Ask for 10 possible book outlines that match the chosen features, randomly pick between 2-8.

5. Create a detailed prompt that includes all the above features, and ask for a synopsis for each chapter, given the above outline chosen.

6. Given {features} and {outline} and {synopsis} write chapter 1.

7. for each chapter in list, given {...} and (optional) previous matching chapter(s), write chapter n+1

(optional 8.) given {...} and 2-3 consecutive chapters, align the ending / beginning of a new chapter for style / features / continuity, etc.

(optional 9.) given {...} and the whole book, list chapters / paragraphs that don't match the given {...} and provide a list of 5 improvements. (randomly choose 1 and ask for an edit).

----

Now, this probably won't give you something like cloud atlas, but they'll at least be different books. That's how I'd do it if I wanted to see how different they can write. Not 1000 "basic" prompts and expecting originality.

Re: The 100k whys of AI

#46

Post by noduerme »

vintermann wrote: post #37
noduerme wrote: post #30

I think a majority of content consumers can already distinguish LLM content from human content. I'm looking forward to the day that they're intelligent enough to care, but I'm not holding my breath. Orwell framed it pretty well in 1984 with the machine-generated songs that were new every year, but always tugged on the heartstrings of the proles. They weren't really readers or listeners to music or appreciators of art…

It's not an "already", because I assume models will get better at addressing mode collapse. The irony in the machine generated songs in 1984 was that Winston clearly found meaning in them, feeling like they applied to him, even though he knew they were machine generated: (from memory) "Under the shade of the chestnut tree / I sold you and you sold me / here lie they and here lie we / under the shade of the chestnut t…

I'm not sure the Chestnut Tree song was machine generated. [edit: I also recall Winston thinking that the proles songs were sappy and repetitive]. I took that as an older song predating the machine slop. But maybe you're right, and if so it's a sadder and deeper irony.

Finding meaning in slop is not ennobling of the human spirit, and I see no reason to champion it.

Also if the meaning is that I sold you and you sold me; what is the upside here?

Re: The 100k whys of AI

#47

Post by cryo32 »

ekianjo wrote: post #34
dlenski wrote: post #2

A nice illustration of the homogeneity of LLM responses. Another way to describe this effect would be… If you ask humans to write 1,000 books, you're asking 1,000 different humans with different experiences and different skills and different moods (etc.) to write those books. But if you ask LLMs to write 1,000 books, you're probably only talking to 3 or 5 different models, tops. And they've all trained on the same or…

prompts will give very different results. this is where you do the work.

I disagree. The LLM outputs really do lack anything original or interesting. They just produce banal copy whatever you ask them.

A good editor could probably reduce all LLM outputs on a subject down to the same point.

Re: The 100k whys of AI

#48

Post by noduerme »

NitpickLawyer wrote: post #45
dlenski wrote: post #2

A nice illustration of the homogeneity of LLM responses. Another way to describe this effect would be… If you ask humans to write 1,000 books, you're asking 1,000 different humans with different experiences and different skills and different moods (etc.) to write those books. But if you ask LLMs to write 1,000 books, you're probably only talking to 3 or 5 different models, tops. And they've all trained on the same or…

> A nice illustration of the homogeneity of LLM responses. [...] And they've all trained on the same or similar data, and are trained to respond in very similar ways. I mostly agree, but this is a very simplified explanation. The models are indeed trained to respond in similar ways, for "basic" prompts. And that's as much a feature as it is a bug. In other words, the bug becomes apparent only if you give 100+ basic p…

That whole thing would get you 1000 variants of existing art. But if you asked a thousand different designers to do a cover for the same book...

Re: The 100k whys of AI

#49

Post by roenxi »

supriyo-biswas wrote: post #35
roenxi wrote: post #9

I don't know how much of a smoking gun this actually is, the evidence proffered doesn't establish anything - I can see some names there like Havilah Brooks or Celina Briar who are intentionally re-using the same title to create a series, for example. And this doesn't really get into the base rate of generic title re-use among encyclopedias. There isn't much reward for coming up with an imaginative title for kids, the…

Have you seen the content of the books in the tweet[1] linked below the article? Between horses with fused butts and other diagrams that don't say as much as they purport to, the cover is the least of its problems, although the only one that can be criticized directly. [1] https://infosec.exchange/@lcamtuf/116785283147249092

This seems to be some out of context pictures where I have no idea what they are meant to be showing or whether they succeeded. And although the cats and zebras are clear AI images again that doesn't mean anything, being exactly 2 pictures presented with no context apart from being seen in a book. So there is a book where the editor was lazy and let some bad AI images through.

I'm sure someone deeply familiar with childrens publishing would be able to talk authoritatively on the extent of new trends, but this seems to be the infosec community and the evidence offered doesn't seem to actually be evidence of anything. There isn't a baseline. Children's encyclopedias might have been a hard-hitting game of radical creativity and high standards in the past, or it could be an endless tide of derivative swill.

And using AI images seems unrelated. That's something people should just be doing. Ideally with better proofreading, but hey. The article's complaint was about lack of originality.

Re: The 100k whys of AI

#50

Post by NitpickLawyer »

noduerme wrote: post #48
NitpickLawyer wrote: post #45

Earlier quoted context omitted.

> A nice illustration of the homogeneity of LLM responses. [...] And they've all trained on the same or similar data, and are trained to respond in very similar ways. I mostly agree, but this is a very simplified explanation. The models are indeed trained to respond in similar ways, for "basic" prompts. And that's as much a feature as it is a bug. In other words, the bug becomes apparent only if you give 100+ basic p…

That whole thing would get you 1000 variants of existing art. But if you asked a thousand different designers to do a cover for the same book...

> 1000 variants of existing art.

This is very naive. I can almost guarantee that some combinations of 20 * 50 features will hit on something that has never been written before in that specific combination. And if that's still not enough, increase the number of features. Add more randomness, add more steering, add random steering in random chapters, change it up, and so on.

Post reply on HN