Hacker News Reader

Hacker News

Comment by subroutine | Hacker News Reader

Parent Full thread

subroutine

The modern news website: 15 sentences, 6 Shutterstock photos, 35 ads

View on HN

Don't forget:

> This article is made and published by Anna Hartz, which may have used AI in the preparation

The editors don't even know for sure if the author used AI or not.

I would put the emphasis on a different word:

> This article is made and published by Anna Hartz, which may have used AI in the preparation

Which, not who. They're not even sure the author is human!

Might be a language issue, as English is not the primary language of the newspaper's staff.

I don't understand the snark.

Considering that "using AI" can mean anything from "AI wrote the whole article" to "the author used AI to check the grammar", I'd argue this disclaimer is unnecessary and it's safe to assume AI is involved in some way nowadays.

(the author of this comment may have used AI)

> it's safe to assume AI is involved in some way nowadays

I don't think it's safe to assume so at all. Granted, I only know one journalist, and they've told me they only use LLMs in their work to gather further sources/references to check, everything else they still do "manually" with their own hands.

The editorial team should know exactly the scope of their teams AI usage. The snark mostly comes from them not knowing if AI was used or not, and they be upfront about them not knowing it. Feels like they're missing integrity if they don't know such things.

> I don't think it's safe to assume so at all. Granted, I only know one journalist, and they've told me they only use LLMs in their work to gather further sources/references to check, everything else they still do "manually" with their own hands.

I'd argue that your example falls under "which may have used AI in the preparation", which was exactly my point. (I actually had using AI for research as an example, but English is not my first language and I couldn't get the sentence to sound correct and chatGPT suggested I drop it)

> The editorial team should know exactly the scope of their teams AI usage. The snark mostly comes from them not knowing if AI was used or not, and they be upfront about them not knowing it. Feels like they're missing integrity if they don't know such things.

I don't see this as a lack of integrity, but rather as a futile attempt at being transparent. Everyone else is in the same position, they are just not adding a disclaimer.

And that's nothing specific about journalists, this applies to all professions. At most you can say what your official policy states, but you have absolutely no way of knowing how your employees/coworkers are using AIs.

> chatGPT suggested I drop it

AI suggests you drop disclosing possible usage of AI.

If it was smart we'd say this was AI influencing the narrative ;)

By using the word "which" instead of "who", the editors don't even know for sure if the author _is_ AI or not.

TBH I'm not even sure if I'm using AI half the time. I'm pretty sure any sort of autocomplete gets marketed as AI these days, so anything I type with my phone probably counts as having used AI. "AI" is a marketing term with no fixed technical meaning.

I'm well aware that AI is very broad, but for all intents and purposes, when journalists say something like "AI was/wasn't used to write this article" they're not talking about autocomplete or being affected by A/B tests that happen to use ML, it's pretty clear they're using AI as another name for LLMs.

That aspect I don’t find all that concerning. I just wish they had a page where they explain what AI preparation entails.

But they can't. That would require resources which they have outsourced to OpenAI. If they documented their workflow all "productivity" gains would be lost. The whole point of AI is to get away with cheap and dirty. sounding high quality and competent.

God forbid the management UI of the publication adds one or two checkboxes with "Did you AI to prepare for this article?" and "Did you use AI to write any of the text for this article?" that authors could just check/uncheck themselves.

"AI" isn't a well-defined term, it's a marketing term. AI includes LLMs, but also seems to include basic autocomplete, spellcheck, and any other algorithm that runs on a computer.

> may

They don't even specifically know for each article if their authors use LLMs or not. What a shitshow.

It's as if they're trying to mimic a Twitter post thread, fearing that people won't have the attention span to read a regular article. This is the equivalent to jump cuts in videos.

The news industry: "Google are killing us! Give us money!"

Well, it's a bit off-topic, but yes, Google is demonstrably killing the fabric of the Internet. Ever since they introduced the info box & knowledge graph, pretty much every change they've introduced is geared towards one goal: they don't want to just be a guide, they want to be the journey AND the destination. All the latest AI efforts are the apotheosis, the extreme logical conclusion of that. Surely I don't need to explain how this makes them extremely similar to a parasite killing its host, how this degrades the very ecosystem their foundations are built on?

? Yes? That's exactly what's happening, google's and meta's algos dictate which sites get traffic which is why every website is slowly but surely turning into clickbait/reels.

At least we don't see the ads.

Reply on news.ycombinator.com