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

I know you didn't write this

ammil.industries

Post reply on HN
Print view View original

91–100 of 145 posts

Re: I know you didn't write this

#91

Post by poemxo »

skwirl wrote: post #59
a1j9o94 wrote: post #36

I know I'm an outlier on HN, but I really don't care if AI was used to write something I'm reading. I just care whether or not the ideas are good and clear. And if we're talking about work output 99% of what people were putting out before AI wasn't particularly good. And in my genuine experience AI's output is better than things people I worked with would spend hours and days on. I feel like more time is wasted tryin…

>But it shouldn't matter if he gave 5 bullets to Chat gpt that expanded it to a full page with a detailed plan. The coworker should just give me the five bullet points they put into ChatGPT. I can trivially dump it into ChatGPT or any other LLM myself to turn it into a "plan."

How do you know the coworker didn't bully the LLM for 20 minutes to get the desired output? It isn't often trivial to one-shot a task unless it's very basic and you don't care about details.

Asking for the prompt is also far more hostile than your coworker providing LLM-assisted word docs.

Re: I know you didn't write this

#92

Post by zephen »

embedding-shape wrote: post #28
zephen wrote: post #24

Earlier quoted context omitted.

> I don't really care if it's a person or the LLM getting it wrong, You're nicer than some of us. If it's an LLM getting it wrong, and it's not caught before it gets to you, then what value is the intermediary adding to the process?

I meant it in a way that I'll blame the person regardless of who actually wrote the text. The LLM messed up and you failed to notice? I blame you. You messed up and failed to notice? I still blame you.

I get that.

But, as discussed in some other threads, the leverage provided by the LLM allows the miscreant to inundate you with slop by only pressing a few buttons.

And rejection is work. So they can produce more slop, requiring more rejections, faster than you can read the slop.

This is what's new. You reject it, they feed your rejection back into the LLM, and hand you something 5 minutes later with so many formatting changes that diff is unhelpful, and enough subtle substantive changes embedded in it that if you don't read the entire thing, you might have missed something important.

Re: I know you didn't write this

#93

Post by patrickmay »

macrael wrote: post #51

I think it quickly needs to become good manners to indicate when text was written by AI rather than a person. I read that text differently and I shouldn't have to spend my time guessing.

A footnote with the prompt used would be even more polite. Then I can just read that and skip the generated text.

Re: I know you didn't write this

#94

Post by teaearlgraycold »

isodev wrote: post #81

Once, I had a very frustrating slack chat with a fellow developer. We were discussing edge cases for a new feature and the experience from my perspective was that for each of my messages, I’d get a “in case of … how about …” style reply. The topic was focused around iOS vs. Android app lifecycle. Every now and then my colleague would suggest APIs or events that simply don’t exist. This was before vibe coding, around…

Anyone doing this should be fired. Both for the lack of trust they bring to the team but also because they’re just making themselves a middle man to an LLM. Why not cut out the middle man?

Re: I know you didn't write this

#95

Post by amarant »

a1j9o94 wrote: post #36

I know I'm an outlier on HN, but I really don't care if AI was used to write something I'm reading. I just care whether or not the ideas are good and clear. And if we're talking about work output 99% of what people were putting out before AI wasn't particularly good. And in my genuine experience AI's output is better than things people I worked with would spend hours and days on. I feel like more time is wasted tryin…

Agreed! I've reached the conclusion that a lot of people have completely misunderstood why we work.

It's all about the utility provided. That's the only thing that matters in the end.

Some people seem to think work is an exchange of suffering for money, and omg some colleagues are not suffering as much as they're supposed to!

The plan(or any other document) has to be judged on its own merits. Always. It doesn't matter how it was written. It really doesn't.

Does that mean AI usage can never be problematic? Of course not! If a colleague feeds their tasks to a LLM and never does anything to verify quality, and frequently submits poor quality documents for colleagues to verify and correct, that's obviously bad. But think about it: a colleague who submits poor quality work is problematic regardless of if they wrote it themselves or if they had an AI do it.

A good document is a good document. And a bad one is a bad one. Doesn't matter if it was written using vim, Emacs or Gemini 3

Re: I know you didn't write this

#96

Post by zephen »

EGreg wrote: post #65

I think that we should have revision control for intermediate stages - for code, documents, even paintings. So we can at least have some idea of provenance, how it's made. Until AI is used to fake that, too.

There are some issues with this.

1) For things made with LLMs: 1a) The fact that older versions aren't online forever. You literally might never be able to put the original prompt in and get the same result. 1b) A minor change in input prompt can result in a huge output change, rendering the original prompt practically meaningless, especially if modifications were required for the output of the LLM.

2) For things made the old-fashioned way, most history is boring and not useful. The best git repos have carefully curated history, with cohesive change sets that are both readable, and usable when bisecting the commit history for regressions.

Re: I know you didn't write this

#97

Post by kazinator »

Plans usually start a short lists of ideas without a lot of detail. When people discuss and agree on things, then choices are decided upon. The branches of the "search tree" which are not being taken are pruned away and detail is added to the path taken.

If someone just generates an incredibly detailed plan in one go, that destroys the process. Others now are wasting time looking at details in something that may not even be a good idea if you step back.

The successive refinement flow doesn't preclude consideration of input from AI.

Re: I know you didn't write this

#98

Post by pgwhalen »

GaryBluto wrote: post #3
messe wrote: post #2

> Suspicions aroused, I clicked on the “Document History” button in the top right and saw a clean history of empty document – and then wham – fully-formed plan, as if it had just spilled out of someone’s brain, straight onto the screen, ready to share. This isn't always a great indicator. I can't stand Google Docs as an interface to write with, so use VIM and the copy/paste the completed document into it.

It's bizarre to me that this didn't occur even slightly to the post author.

It probably did, but they didn't feel the need to fully explain why they were confident it was AI generated, since that's not the point of the article.

Re: I know you didn't write this

#100

Post by zephen »

teeray wrote: post #29

> So it’s definitely AI. I felt betrayed and a little foolish. But why? Because the prompter is basically gaslighting reviewers into doing work for them. They put their marks of authorship on the AI slop when they've barely looked at it at all which convinces the reviewer to look. When the comments come back, they pump the feedback into the LLM, more slop falls out and around we go again. The prompter isn't really do…

Not sure why this is being downvoted. It accurately and succinctly describes a likely reason for a _feeling_.
Post reply on HN