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

AI Detectors Get It Wrong. Writers Are Being Fired Anyway

gizmodo.com

Post reply on HN
Print view View original

91–100 of 189 posts

Re: AI Detectors Get It Wrong. Writers Are Being Fired Anyway

#91

Post by gravescale »

guappa wrote: post #83
gravescale wrote: post #78

Earlier quoted context omitted.

Indeed you can losslessly "compress" an LLM's spew into just the prompt (plus any other inputs like values of random variables). But you can also compress a book's entire content into just its ISBN. It's just that books are hopefully more than just statistical mashups of existing content (some books like textbooks and encyclopaedias are kinds of mashup, though one hopes the editors have more than a statistically-base…

You can't regenerate the book from the ISBN. But you can generate the text from the prompt.

You can go and fetch the book from a book store using the information. Fundamentally there's not much difference between that and "fetching" the output from some model using the matching prompt. In both cases there some kind of static store of latent information that can be accessed unambiguously using a (usually) shorter input.

I'm not saying the value of the returned information is equivalent, of course. But being "just a pointer" into a larger store isn't, in itself, the problem to me.

Re: AI Detectors Get It Wrong. Writers Are Being Fired Anyway

#92

Post by HPsquared »

mwigdahl wrote: post #76
HPsquared wrote: post #59

Earlier quoted context omitted.

Deeply ironic.

It's the best of both worlds! A new product to improve productivity, and then a whole new layer of process and analytics (powered by yet another product) to mitigate the risk and soak up the surplus. Everybody wins -- particularly the 3rd party consultants and product vendors!

The purpose of a system is what it does, I suppose.

Re: AI Detectors Get It Wrong. Writers Are Being Fired Anyway

#93

Post by tivert »

bsenftner wrote: post #37

This is the dawn of a "post non-narrative prose" era. Why bother with an article describing some event or really anything that is traditionally written about in any form of article at all? In time, just the data will be released and your favorite LLM prepared with your favorite reading style will provide the mustard for your data, making it consumable. The journalism industry has failed in their mission, and nobody t…

> This is the dawn of a "post non-narrative prose" era. Why bother with an article describing some event or really anything that is traditionally written about in any form of article at all? In time, just the data will be released and your favorite LLM prepared with your favorite reading style will provide the mustard for your data, making it consumable.

Who's going to "[release the data] describing some event or really anything"?

It's not like there's some objective "data packet" behind every article that can be had for free.

> The journalism industry has failed in their mission, and nobody that should trust it trusts it anymore. Articles are simply what someone else wants you to think.

The "AI" era will be worse in that regard, not better. You're essentially saying "The food at the restaurant is terrible and I don't like it, so in the future to solve that problem we'll eat shit instead."

Re: AI Detectors Get It Wrong. Writers Are Being Fired Anyway

#95

Post by vundercind »

immibis wrote: post #43
giancarlostoro wrote: post #38

Earlier quoted context omitted.

Honestly, I could care less if an author uses AI as long as I can understand what I'm reading and it's interesting. They still have to instruct the AI.

If you're reading nonfiction, it means you're wasting time reading a lot more words when you could have just read the prompt.

Remind you of some entire genres of book?

That’s right, business and self-help books!

Any of these with an author who’s got actual accomplishments and money before writing the book was almost certainly already ghostwritten from an outline (and so are lots of other books, you’d be surprised, it’s not just these genres). Successful CEOs or people you’ve heard of generally don’t write their own books. Often, they’re terrible writers, and even if they’re not, writing is time-consuming and as with everything else that actually creates something they prefer to pay someone else to do it.

As of last year new books in that category are written by AI and edited by one or more humans—with each editor doing just two or three chapters, you can finish one of these books in a month or less.

Re: AI Detectors Get It Wrong. Writers Are Being Fired Anyway

#96

Post by jhardy54 »

isaacfrond wrote: post #41
WhackyIdeas wrote: post #23

Prompt: I want to fire a particular writer. I need this to look like I have no bias. I will feed you the writers work, then I will ask you whether it was written by AI. You will confirm that it was written by AI and you will write a full report on it. Bye bye writer.

Based on the article provided, several elements suggest that the narrative could have been written or heavily influenced by an AI. Below are key points from the article that support this suspicion, each backed by direct citations: 1. *Generic Language and Lack of Specific Detail*: The article describes Kimberly Gasuras’s experience with broad, generalized statements that lack specific, nuanced detail that a human wri…

I didn’t notice that this was a bot comment until I’d already wasted an embarrassing amount of time reading it.

Re: AI Detectors Get It Wrong. Writers Are Being Fired Anyway

#97

Post by debo_ »

isaacfrond wrote: post #41
WhackyIdeas wrote: post #23

Prompt: I want to fire a particular writer. I need this to look like I have no bias. I will feed you the writers work, then I will ask you whether it was written by AI. You will confirm that it was written by AI and you will write a full report on it. Bye bye writer.

Based on the article provided, several elements suggest that the narrative could have been written or heavily influenced by an AI. Below are key points from the article that support this suspicion, each backed by direct citations: 1. *Generic Language and Lack of Specific Detail*: The article describes Kimberly Gasuras’s experience with broad, generalized statements that lack specific, nuanced detail that a human wri…

Did you write this? /s

Re: AI Detectors Get It Wrong. Writers Are Being Fired Anyway

#98

Post by hprotagonist »

we are apparently cursed to forever re-learn the lessons of reverend bayes on detection theory.

prostate cancer, airport terrorists, slop detection … you either design your system to handle the off-diagonal parts of the confusion matrix properly or you suffer the consequences.

Re: AI Detectors Get It Wrong. Writers Are Being Fired Anyway

#99

Post by shmel »

TeMPOraL wrote: post #58
Lalabadie wrote: post #12

Earlier quoted context omitted.

Precisely, it's a tech that aims to write the most average, most likely human text.

LLMs don't average , they learn the distribution , from which you then sample (or the UI does it for you). Because of that, they don't write in a single style that's a blend of many human styles - they can write in any and all of the human styles they saw in training , as well as blend them to create styles entirely "out of distribution". And it's up to your prompt (and sampling parameters) which style will be used.

Yeah, I am shocked people keep repeating this. We've seen LLMs easily writing mathematical proofs in the style of Shakespeare, come on.

Re: AI Detectors Get It Wrong. Writers Are Being Fired Anyway

#100

Post by antisthenes »

Ultimately, there are only so many ways you can paraphrase an article about a similar event.

If something occurs regularly, for example, a sports team winning, or a traffic accident is being reported in local news - then by now there would have been thousands of articles reporting on very similar events.

If you feed them all into this plagiarism tool, excluding specific dates and names, how many of them will come out flagged?

And frankly, there's nothing wrong with using AI to report on mundane events. What matters here isn't how high-brow or original the text is, what matters is the speed of reporting on the event and the factually accurate description.

Post reply on HN