Thread

Too AI; Didn't Read

111 points · 113 comments · rfonseca

lemontheme · 1790370169 · focus · HN ↗
If it’s not worth writing, it’s not worth reading.
That said, wouldn’t it be funny if this site, too, were AI generated?
1. bonoboTP · 1790370880 · focus · HN ↗
  The other similar site making the rounds, <a href="https://noslopgrenade.com/" rel="nofollow">https://noslopgrenade.com/, is indeed in part written by AI as evaluated by Pangram.
  1. nullc · 1790371348 · focus · HN ↗
    I believe that one was discussed with the author on HN, and that they used AI to 'copyedit'. Unfortunately that's a trap because if you make a general copyediting request most LLMs will absolutely slopify your work-- and make edits that are unambiguously bad as well as just making it look like lazy llm output.
    The better use is to ask it to point out grammatical and spelling errors, or phrasing which is unclear.. then go fix yourself (or direct highly targeted fixes that don't let it slopify). Never hand the clanker the pen for your writing unless you're looking to subject everyone to your Honest Take That Recontextualizes Everything.
    AlexCoventry · 1790371910 · focus · HN ↗
    IME, it's best to ask an AI to criticize your work and suggest changes without providing actual rewrites, and make the changes you agree with in your own voice.
  2. lemontheme · 1790371483 · focus · HN ↗
    Is that with or without the slop grenade examples included, because I can see how those might trigger the detector
    bonoboTP · 1790371740 · focus · HN ↗
    Without.
  3. breezybottom · 1790371845 · focus · HN ↗
    AI detectors are useless. I don't know how this forum got so infected with Pangram shills, but at this point they need to be banned.
    E-Reverance · 1790372074 · focus · HN ↗
    Apart from their own data, a lot of people's anecdotal experience is that its pretty accurate, why would they need to banned for that?
    bonoboTP · 1790372396 · focus · HN ↗
    This is an outdated take. It used to be the case, but now detectors have low false positive rates. In other words good precision but not necessarily good recall for detecting AI. So some AI text evades detection, but when it flags something as AI, then it's usually right.
    breezybottom · 1790536702 · focus · HN ↗
    No, it's not. The only way it could is through watermarking, and even then you need a substantial amount of text for that to be reliable.

Open on Hacker News to reply ↗

Unofficial Hacker News client; not affiliated with Y Combinator.