[flagged]
Here is the work process: https://chat.openai.com/share/648d0c85-9661-4096-afcc-f92779...
I'm not a native speaker either, sometimes I re-read some of my own comments and realise I could have phrased them much better, corrected some grammar but I prefer that they sound like me.
Just found it funny, that even with text is often very easy to tell.
I use "akin" relatively frequently, and "perplexing" ocassionally, as well as other such "rare" (?) words. Is english vocabulary beyond the top-1000 most common words or anything above 2 syllables considered a sign of AI these days?
If anything marks it as AI, it's the overall tone and phrasing (too "mechanistic"), not the use of ";", "akin" or "perplexing".
So they were right on the money.
> Has anyone else noticed that?
Now models could be trained adversarially against this GPTZero and you'd have a harder time..
For many of us our "normal online speech" is largely identical to our normal written speech except when writing somewhere where we either are not using a normal sized physical keyboard or where posts need to be severely limited in length.
Maybe that part did not get rephrased.
Also, I use akin and perplexing quite often in my text. The semicolon much more rarely however. But when I write professional text, then yes, I use semicolons too.
God. GPT is going to fucking ruin online discourse.
A lot of generative content is becoming hard to differentiate from human created content and it's also becoming common to see people like yourself assign false positives. I think it's good to be skeptical but as a skeptic myself I know there's downsides to considering everything suspiciously. Unfortunately I don't think we'll ever find a good solution for wanting to know if what we're looking at is AI or not.