To really understand how this stuff can trip you up, and I’m talking about the math behind it, not the confuse thingy or the proof, you have to understand Perplexity first. So, basically, perplexity is like a surprise meter. So, AI, when it’s working its magic/tokens, is essentially a fancy autocomplete mechanism, and it’s trying to guess the next word based on its statistical likelihood. So, essentially, it’s generating text that’s incredibly smooth. Almost too smooth. So smooth, in fact, that it’s problematic. So, humans, on the other hand, are full of contradictions. So, sometimes we’ll use slang, and sometimes we’ll use weird and wacky ways of talking. So, essentially, we’re full of contradictions, and this is essentially high perplexity. So, now, and this is where this stuff can trip people up, is where Burstiness comes in. So, essentially, this is like a rhythm. So, AI, when it’s working its magic, writes like a metronome. So, tick, tick, tick. So, essentially, its sentences are all pretty similar in length and style. So, it’s just this constant, almost boring, rhythm. So, humans, on the other hand, don’t write like this. So, sometimes we’ll write a sentence, and then we’ll write a fragment, and then we’ll write a sentence, and then we’ll write a long, meandering sentence with three clauses because we’re trying to get this messy idea out before we run out of breath.
So yeah......
See how easily?
I don't use AI for debates
Source: Trust me bro
If this comment is detected as AI then Im genuinely f*cked up
Honestly, I copy and pasted this argument but not what I wrote precedently:
How can I trust you won't use AI, we both rely on the same burden of proof
I could have written my argument by myself (which I did) and then could have still responded with AI (to lower suspect)
It does not make sense.
Here is the argument in its most rigorous and cautious form:
Objective moral obligations exist.
(Raping toddlers for fun is really wrong, necessarily wrong, wrong whether anyone believes it or not, and wrong in every possible world in which such an act occurs.)
and etc.
GPTZero relies on perplexity and burstiness, so you can't trust it fully either
You think so?
To really understand how this stuff can trip you up, and I’m talking about the math behind it, not the confuse thingy or the proof, you have to understand Perplexity first. So, basically, perplexity is like a surprise meter. So, AI, when it’s working its magic/tokens, is essentially a fancy autocomplete mechanism, and it’s trying to guess the next word based on its statistical likelihood. So, essentially, it’s generating text that’s incredibly smooth. Almost too smooth. So smooth, in fact, that it’s problematic. So, humans, on the other hand, are full of contradictions. So, sometimes we’ll use slang, and sometimes we’ll use weird and wacky ways of talking. So, essentially, we’re full of contradictions, and this is essentially high perplexity. So, now, and this is where this stuff can trip people up, is where Burstiness comes in. So, essentially, this is like a rhythm. So, AI, when it’s working its magic, writes like a metronome. So, tick, tick, tick. So, essentially, its sentences are all pretty similar in length and style. So, it’s just this constant, almost boring, rhythm. So, humans, on the other hand, don’t write like this. So, sometimes we’ll write a sentence, and then we’ll write a fragment, and then we’ll write a sentence, and then we’ll write a long, meandering sentence with three clauses because we’re trying to get this messy idea out before we run out of breath.
So yeah......
See how easily?
I don't use AI for debates
Source: Trust me bro
If this comment is detected as AI then Im genuinely f*cked up
Honestly, I copy and pasted this argument but not what I wrote precedently:
How can I trust you won't use AI, we both rely on the same burden of proof
I could have written my argument by myself (which I did) and then could have still responded with AI (to lower suspect)
It does not make sense.
Here is the argument in its most rigorous and cautious form:
Objective moral obligations exist.
(Raping toddlers for fun is really wrong, necessarily wrong, wrong whether anyone believes it or not, and wrong in every possible world in which such an act occurs.)
and etc.
GPTZero relies on perplexity and burstiness, so you can't trust it fully either
Clear stance
xD xD