3 ms·

fyi the announcement is very obviously ai-written. It’s off-putting.

by antonvs 9mo ago

moWerk 9mo ago
Fair suspicion in 2026 — but we actually started drafting that announcement in early 2023 before first committing it here: https://github.com/jrtberlin/aos2.0-post https://github.com/jrtberlin/aos2.0-post. As a non native speaker i sure ran my commits through grammar polishing multiple times over the years. And i am genuinely curious now what to avoid to not sound like an LLM if you could dig out one or two examples.
- bjackman 9mo ago
  FWIW as someone who sadly has about 10% of his brain constantly engaged in AI detection when I read HN, absolutely nothing about the announcement struck me as feeling AI generated. (I feel guilty for pointing it out but I guess it's actually a compliment in this context: I even saw a beautifully reassuring mis-conjugated verb elsewhere on the website. I wonder if LLMs will have to start injecting these errors to give us an authentic feel. Maybe they already have). I think we are at a point now where just coz AI is so prevalent, every post on any programming forum will have at least one comment saying "AI slop".
  - antonvs 9mo ago
    Read the text of the HN post. If it’s not AI slop, then it’s a canonical example of the training data that produces the slop.
- Aachen 9mo ago
  I hadn't opened the article yet and was just browsing comments over my cereal when I saw this and thought "ugh, amother one?" and went to check for myself I didn't get an LLM vibe at all. Looking for it specifically, the bullet point about UI improvements is a candidate; the sentence following "mediawiki" could be an autocompletion; maybe the first sentence of the download section... but they're also all plausibly just a bit 'marketing team worded', so not necessarily LLM-sounding. And even if an LLM made suggestions to some small parts like these, who cares? There aren't any slop sections that waste your time, this is just like using a thesaurus — if these parts were LLM suggestions in the first place, which I don't actually expect because then there should be more of it This type of post very poorly lends itself for auto-writing anyway. It'll put emphasis on the wrong aspects and not come out as intended, at least in my experience it's more work coaxing it to good results. It can be helpful not to start from a blank page but that's about it, I rarely find a sentence among the output that's fully usable as-is
  - antonvs 9mo ago
    I was referring to the text of the HN post. Go read it.
    - Aachen 9mo ago
      Aye-aye o7 ... I saw that one already and didn't suspect slop there either
- retsibsi 9mo ago
  I wonder if they meant this HN post (the OP), rather than https://asteroidos.org/news/2-0-release/ https://asteroidos.org/news/2-0-release/? The latter doesn't seem LLM-written to me, but the HN post does.
  - antonvs 9mo ago
    Correct.
    - moWerk 9mo ago
      Full disclosure, i spent ~3 hours crafting this post to hit the tone i wanted to convey and am kind of proud of the wrist-size linux banger i came up with. I am usually not good with writing since it takes me ages and i might be a bit over sensitive right now. All i wanted was to spare you all the rocky grammar as a cherry on top. To now find that the polished version triggers your "its (completely?) Ai written" sensor. Lesson learned i guess.
- antonvs 9mo ago
  I was referring to the text of the HN post. The second and third paragraphs in particular, but most of it, really. If you didn’t use an LLM for that then congratulations, you’ve successfully nailed the Silicon Valley LinkedIn marketing guru zeitgeist which LLMs seem to gravitate towards.
prmoustache 9mo ago
What is off-putting ia this kind of comments everywhere you go. This is the new "this has clearly been photoshoped" meme we used to see on every forum thread 2 decades ago and it is annoying as hell.
- nunobrito 9mo ago
  Not on this article, but everywhere nowadays are hints of AI. For example using emoticons on the beginning of each sentence or never really going deep into the topic. The complete lack of grammatical errors is another tip, unless the prompter asks to write minor errors and pretend to be a human. Wouldn't be suprised that before the end of this year we see more website declaring themselves to be "human-only zones", considering we won't be able to match the speed and quantity and replies from bots. Making it difficult to hold proper conversations.
  - thinkingemote 9mo ago
    This is one theory about the controversial age identification currently being implemented happily by social media companies: it's actually about human identification. That's the new captcha.
  - mapcars 9mo ago
    I don't care if the text is ai-written as long as it delivers the message that it is supposed to deliver correctly. If none of the engineers on the team feels like writing long announcement posts - why couldn't they use ai, check/correct its output and go back enjoy hacking smartwatches? I think people are getting too fixated on how exactly each symbol was created instead of what the message is.
    - nunobrito 9mo ago
      There is misunderstanding. There is nothing against using it for those cases, the problem is when it gets used for producing long and unverified slop or debatting against humans.
- 9mo ago

fyi the announcement is very obviously ai-written. It’s off-putting.

by antonvs 9mo ago

fyi the announcement is very obviously ai-written. It’s off-putting.

Fair suspicion in 2026 — but we actually started drafting that announcement in early 2023 before first committing it here: https://github.com/jrtberlin/aos2.0-post https://github.com/jrtberlin/aos2.0-post. As a non native speaker i sure ran my commits through grammar polishing multiple times over the years. And i am genuinely curious now what to avoid to not sound like an LLM if you could dig out one or two examples.

bjackman 9mo ago
FWIW as someone who sadly has about 10% of his brain constantly engaged in AI detection when I read HN, absolutely nothing about the announcement struck me as feeling AI generated. (I feel guilty for pointing it out but I guess it's actually a compliment in this context: I even saw a beautifully reassuring mis-conjugated verb elsewhere on the website. I wonder if LLMs will have to start injecting these errors to give us an authentic feel. Maybe they already have). I think we are at a point now where just coz AI is so prevalent, every post on any programming forum will have at least one comment saying "AI slop".
- antonvs 9mo ago
  Read the text of the HN post. If it’s not AI slop, then it’s a canonical example of the training data that produces the slop.
Aachen 9mo ago
I hadn't opened the article yet and was just browsing comments over my cereal when I saw this and thought "ugh, amother one?" and went to check for myself I didn't get an LLM vibe at all. Looking for it specifically, the bullet point about UI improvements is a candidate; the sentence following "mediawiki" could be an autocompletion; maybe the first sentence of the download section... but they're also all plausibly just a bit 'marketing team worded', so not necessarily LLM-sounding. And even if an LLM made suggestions to some small parts like these, who cares? There aren't any slop sections that waste your time, this is just like using a thesaurus — if these parts were LLM suggestions in the first place, which I don't actually expect because then there should be more of it This type of post very poorly lends itself for auto-writing anyway. It'll put emphasis on the wrong aspects and not come out as intended, at least in my experience it's more work coaxing it to good results. It can be helpful not to start from a blank page but that's about it, I rarely find a sentence among the output that's fully usable as-is
- antonvs 9mo ago
  I was referring to the text of the HN post. Go read it.
  - Aachen 9mo ago
    Aye-aye o7 ... I saw that one already and didn't suspect slop there either
retsibsi 9mo ago
I wonder if they meant this HN post (the OP), rather than https://asteroidos.org/news/2-0-release/ https://asteroidos.org/news/2-0-release/? The latter doesn't seem LLM-written to me, but the HN post does.
- antonvs 9mo ago
  Correct.
  - moWerk 9mo ago
    Full disclosure, i spent ~3 hours crafting this post to hit the tone i wanted to convey and am kind of proud of the wrist-size linux banger i came up with. I am usually not good with writing since it takes me ages and i might be a bit over sensitive right now. All i wanted was to spare you all the rocky grammar as a cherry on top. To now find that the polished version triggers your "its (completely?) Ai written" sensor. Lesson learned i guess.
antonvs 9mo ago
I was referring to the text of the HN post. The second and third paragraphs in particular, but most of it, really. If you didn’t use an LLM for that then congratulations, you’ve successfully nailed the Silicon Valley LinkedIn marketing guru zeitgeist which LLMs seem to gravitate towards.

prmoustache 9mo ago

What is off-putting ia this kind of comments everywhere you go. This is the new "this has clearly been photoshoped" meme we used to see on every forum thread 2 decades ago and it is annoying as hell.

nunobrito 9mo ago
Not on this article, but everywhere nowadays are hints of AI. For example using emoticons on the beginning of each sentence or never really going deep into the topic. The complete lack of grammatical errors is another tip, unless the prompter asks to write minor errors and pretend to be a human. Wouldn't be suprised that before the end of this year we see more website declaring themselves to be "human-only zones", considering we won't be able to match the speed and quantity and replies from bots. Making it difficult to hold proper conversations.
- thinkingemote 9mo ago
  This is one theory about the controversial age identification currently being implemented happily by social media companies: it's actually about human identification. That's the new captcha.
- mapcars 9mo ago
  I don't care if the text is ai-written as long as it delivers the message that it is supposed to deliver correctly. If none of the engineers on the team feels like writing long announcement posts - why couldn't they use ai, check/correct its output and go back enjoy hacking smartwatches? I think people are getting too fixated on how exactly each symbol was created instead of what the message is.
  - nunobrito 9mo ago
    There is misunderstanding. There is nothing against using it for those cases, the problem is when it gets used for producing long and unverified slop or debatting against humans.
9mo ago