[precompute]

I don't think it was written by a LLM, the words have a warmth that generated text lacks.

Plus you can tell it's written as if it's half-spoken.

[dmd]

> warmth

That shows the LLM is running on vacuum tubes.

[fxtentacle]

Perfect. My next business venture is selling ChatGPT text-warming equipment to audiophile readers.

[kergonath]

And gold-plated pure monocrystalline copper connectors.

[iLoveOncall]

> the words have a warmth that generated text lacks

Reminds me of my philosophy teacher in high-school that gave me the best grade of the year and praised how "personal" it felt, for an essay I had bought online for 3 euros.

The irony of it still makes me chuckle 15 years later.

[precompute]

Where's the irony?

[LeonB]

The lecturer perceived it to be “personal” implying authentic and genuine, but in truth the essay was the opposite of these things, that is ironic.

[precompute]

Sure, but it was written by a human and not a LLM.

[m348e912]

I think being able to identify LLM generated text would be a nice skill to have, but unfortunately I don't think it's as easy as you think it is.