“it’s not x, it’s y”, forcing the rule of 3 into every paragraph, flowery language that doesn’t add anything or make much sense (e.g.,“it’s the silence in your chest where the certainty used to be”) and other tells that this was AI generated. Do you not notice that this is not how humans speak?
The rule of three and “it’s not x, it’s y” might not be how humans typically speak, but it is how humans write. Some of us have even been known to use the Oxford comma. That’s why AI reproduces it.
(Seriously, you should use the Oxford comma. That shit decides legal cases.)
“It’s the silence in your chest where the certainty used to be” does sound like AI bullshit, though.
Meh, I used to use almost all the same methods, including the em dashes, ending conclusions with a mic drop statement, etc., but LLMs are so specifically patterned that you can spot it a mile away. ChatGPT talks like people, people do not talk like chatGPT.
At the same time, people on the internet don’t seem to be talking this way. Every AI-generated response I have ever gotten looks exactly like that, though.
I’m not basing my assumption on how I talk so the psychology lesson doesn’t quite apply.
“it’s not x, it’s y”, forcing the rule of 3 into every paragraph, flowery language that doesn’t add anything or make much sense (e.g.,“it’s the silence in your chest where the certainty used to be”) and other tells that this was AI generated. Do you not notice that this is not how humans speak?
The rule of three and “it’s not x, it’s y” might not be how humans typically speak, but it is how humans write. Some of us have even been known to use the Oxford comma. That’s why AI reproduces it.
(Seriously, you should use the Oxford comma. That shit decides legal cases.)
“It’s the silence in your chest where the certainty used to be” does sound like AI bullshit, though.
I recall one of the first things being taught in psychology “You are not the norm. What is normal for you, doesnt have to be normal for others.”
It can be how humans speak and it absolutely is how I often speak.
Meh, I used to use almost all the same methods, including the em dashes, ending conclusions with a mic drop statement, etc., but LLMs are so specifically patterned that you can spot it a mile away. ChatGPT talks like people, people do not talk like chatGPT.
Sure, I could be wrong.
At the same time, people on the internet don’t seem to be talking this way. Every AI-generated response I have ever gotten looks exactly like that, though.
I’m not basing my assumption on how I talk so the psychology lesson doesn’t quite apply.
Fair enough. Im at fault for my assumption then. Thanks for clarifying.
I don’t think most LLMs would use “spreadsheeted” as a verb.
That’s exactly the kind of language I would expect it to use.
Seeing as it’s not a word, no.