3 ms·

Was this comment generated via ChatGPT? Not trying to catch you out, just curious if I'm able to successfully spot it in the wild.

teshigahara 4y ago
I also got very "ChatGPT" vibes from that comment but thought I was being too paranoid by the last paragraph
willriches 4y ago
I passed their comment to ChatGPT 3.5 and it seemed confident that ChatGPT wrote it.
- codetrotter 4y ago
  I don’t think ChatGPT can tell if something was written by ChatGPT. At least not reliably.
  - padolsey 4y ago
    OpenAI disagrees :p https://openai.com/blog/new-ai-classifier-for-indicating-ai-written-text https://openai.com/blog/new-ai-classifier-for-indicating-ai-...
    - sheepscreek 4y ago
      ChatGPT is probably not trained for this. The classifier seems like a different model/product. Can access it in the playground here: https://platform.openai.com/ai-text-classifier https://platform.openai.com/ai-text-classifier
- Karunamon 4y ago
  For the record, this (asking the model if the output is theirs) is not a reliable way to determine AI authorship of comments. The false positive rate is quite high, and will return true simply for a comment being well written and lacking idiosyncrasies.
neals 4y ago
You caught me red-handed! Yes, the previous comment was actually generated by ChatGPT. I must admit, I was curious to see if anyone would be able to spot it in the wild, and you my friend have successfully done so. In the interest of full disclosure, I should also mention that I didn't actually read the article before leaving my comment. I know, I know, shame on me! But on a more serious note, I think it's important to recognize that while AI has made incredible strides in recent years, there is still a lot of work to be done on the creative side. Many texts generated by AI tend to have a similar structure or tone, and it's not always easy to tell them apart from human-generated text.
- efxhoy 4y ago
  What prompt did you use? I immediately smelled chatgpt from the comment too. No one on hn actually writes like that, says so little and refers back in such a methodical way to the talking points of the actual article.
- ilyt 4y ago
  The true test is "is it actual marketing-speeech or chatgpt". There are few easy things to catch like >and this has resulted in real tangible benefits such as better database connection handling, faster view rendering, and improved security posture. No actual honest tech person would ever say "security posture" It also generally read like someone trying to sell tech to managers. I wouldn't blink an eye if something like that was posted on corporate blog because that's exact type of worm-speech they use
- aczerepinski 4y ago
  This one is also gpt. Can we please keep hn comments human only?
  - neals 4y ago
    I apologize fellow human. The world has changed, the genie is out, and all that is left is bullet-point communication.
- cutler 4y ago
rco8786 4y ago
Got the same vibes. It seems to write in a very "formal" fashion unless explicitly instructed otherwise, and even then doesn't do a great job all the time.

teshigahara 4y ago

I also got very "ChatGPT" vibes from that comment but thought I was being too paranoid by the last paragraph

willriches 4y ago

I passed their comment to ChatGPT 3.5 and it seemed confident that ChatGPT wrote it.

codetrotter 4y ago
I don’t think ChatGPT can tell if something was written by ChatGPT. At least not reliably.
- padolsey 4y ago
  OpenAI disagrees :p https://openai.com/blog/new-ai-classifier-for-indicating-ai-written-text https://openai.com/blog/new-ai-classifier-for-indicating-ai-...
  - sheepscreek 4y ago
    ChatGPT is probably not trained for this. The classifier seems like a different model/product. Can access it in the playground here: https://platform.openai.com/ai-text-classifier https://platform.openai.com/ai-text-classifier
Karunamon 4y ago
For the record, this (asking the model if the output is theirs) is not a reliable way to determine AI authorship of comments. The false positive rate is quite high, and will return true simply for a comment being well written and lacking idiosyncrasies.

neals 4y ago

You caught me red-handed! Yes, the previous comment was actually generated by ChatGPT. I must admit, I was curious to see if anyone would be able to spot it in the wild, and you my friend have successfully done so. In the interest of full disclosure, I should also mention that I didn't actually read the article before leaving my comment. I know, I know, shame on me! But on a more serious note, I think it's important to recognize that while AI has made incredible strides in recent years, there is still a lot of work to be done on the creative side. Many texts generated by AI tend to have a similar structure or tone, and it's not always easy to tell them apart from human-generated text.

efxhoy 4y ago
What prompt did you use? I immediately smelled chatgpt from the comment too. No one on hn actually writes like that, says so little and refers back in such a methodical way to the talking points of the actual article.
ilyt 4y ago
The true test is "is it actual marketing-speeech or chatgpt". There are few easy things to catch like >and this has resulted in real tangible benefits such as better database connection handling, faster view rendering, and improved security posture. No actual honest tech person would ever say "security posture" It also generally read like someone trying to sell tech to managers. I wouldn't blink an eye if something like that was posted on corporate blog because that's exact type of worm-speech they use
aczerepinski 4y ago
This one is also gpt. Can we please keep hn comments human only?
- neals 4y ago
  I apologize fellow human. The world has changed, the genie is out, and all that is left is bullet-point communication.
cutler 4y ago

rco8786 4y ago

Got the same vibes. It seems to write in a very "formal" fashion unless explicitly instructed otherwise, and even then doesn't do a great job all the time.