18-02-2026, 01:04 AM
This is where I think that judging solution proposals for LLM / AI influence becomes tricky.
To me this looks hybrid, even when the written text screems out LLM. (Not because of the em dashes).
LLM's have long moved away from simple substitution approaches and seem to have 'understood' that Eva is not the meaning of the characters. They tend to take the procedural/algorthmic approach.
I imagine that the Eva/substitution concept could in principle be 'forced upon' the LLM by an interrogator.
Just guessing.
On a side note: I think we already had a Friulian proposal before, and these dialectical features are grossly insufficient to explain low bigram entropy.
To me this looks hybrid, even when the written text screems out LLM. (Not because of the em dashes).
LLM's have long moved away from simple substitution approaches and seem to have 'understood' that Eva is not the meaning of the characters. They tend to take the procedural/algorthmic approach.
I imagine that the Eva/substitution concept could in principle be 'forced upon' the LLM by an interrogator.
Just guessing.
On a side note: I think we already had a Friulian proposal before, and these dialectical features are grossly insufficient to explain low bigram entropy.