|
|
|
|
@@ -1,7 +1,6 @@
|
|
|
|
|
# LLM Prose Tells
|
|
|
|
|
|
|
|
|
|
A catalog of structural, lexical, and rhetorical patterns found in LLM-generated
|
|
|
|
|
prose.
|
|
|
|
|
A catalog of patterns found in LLM-generated prose.
|
|
|
|
|
|
|
|
|
|
---
|
|
|
|
|
|
|
|
|
|
@@ -18,7 +17,7 @@ A negation followed by an em-dash and a reframe.
|
|
|
|
|
|
|
|
|
|
Even outside the "not X but Y" pivot, models substitute em-dashes for commas,
|
|
|
|
|
semicolons, parentheses, colons, and periods. The em-dash can replace any other
|
|
|
|
|
punctuation mark, and models default to it for that reason.
|
|
|
|
|
punctuation mark, so models default to it.
|
|
|
|
|
|
|
|
|
|
### The Colon Elaboration
|
|
|
|
|
|
|
|
|
|
@@ -79,8 +78,7 @@ zero information. The actual point is always in the next paragraph.
|
|
|
|
|
|
|
|
|
|
> "This is, of course, a simplification." "There are, to be fair, exceptions."
|
|
|
|
|
|
|
|
|
|
Parenthetical asides inserted to perform nuance without ever changing the
|
|
|
|
|
argument.
|
|
|
|
|
Parenthetical asides inserted to perform nuance without changing the argument.
|
|
|
|
|
|
|
|
|
|
### The Unnecessary Contrast
|
|
|
|
|
|
|
|
|
|
@@ -127,10 +125,10 @@ precedent), "navigate," "foster," "underscores," "resonates," "embark,"
|
|
|
|
|
|
|
|
|
|
### Elevated Register Drift
|
|
|
|
|
|
|
|
|
|
Models write one register above where a human would. "Use" becomes "utilize."
|
|
|
|
|
"Start" becomes "commence." "Help" becomes "facilitate." "Show" becomes
|
|
|
|
|
"demonstrate." "Try" becomes "endeavor." "Change" becomes "transform." "Make"
|
|
|
|
|
becomes "craft."
|
|
|
|
|
Models write one register above where a human would, replacing "use" with
|
|
|
|
|
"utilize," "start" with "commence," "help" with "facilitate," "show" with
|
|
|
|
|
"demonstrate," "try" with "endeavor," "change" with "transform," and "make" with
|
|
|
|
|
"craft."
|
|
|
|
|
|
|
|
|
|
### Filler Adverbs
|
|
|
|
|
|
|
|
|
|
@@ -307,9 +305,9 @@ introduces another.
|
|
|
|
|
|
|
|
|
|
### Pass 2: Sentence-Level Restructuring
|
|
|
|
|
|
|
|
|
|
10. Find every em-dash pivot ("not X...but Y," "not just X...Y," "more than
|
|
|
|
|
X...Y") and rewrite it as two separate clauses or a single sentence that
|
|
|
|
|
makes the point without the negation-then-correction structure.
|
|
|
|
|
10. Find every em-dash pivot ("not X—but Y," "not just X—Y," "more than X—Y")
|
|
|
|
|
and rewrite it as two separate clauses or a single sentence that makes the
|
|
|
|
|
point without the negation-then-correction structure.
|
|
|
|
|
|
|
|
|
|
11. Find every colon elaboration and check whether it's doing real work. If the
|
|
|
|
|
clause before the colon could be deleted without losing meaning, rewrite the
|
|
|
|
|
@@ -443,12 +441,6 @@ roughly like this:
|
|
|
|
|
>
|
|
|
|
|
> **model:** _(rewrites entire document without em-dashes while describing
|
|
|
|
|
> em-dash overuse)_
|
|
|
|
|
>
|
|
|
|
|
> **human:** this whole document seems to be making the case for FREQUENCY of
|
|
|
|
|
> use being important. we don't care about frequency, remove all that
|
|
|
|
|
> persuasion.
|
|
|
|
|
>
|
|
|
|
|
> **model:** _(strips out every "humans do this too but less often" comparison)_
|
|
|
|
|
|
|
|
|
|
The human compared this process to the deleted scene in Terminator 2 where John
|
|
|
|
|
Connor switches the T-800's CPU to learning mode. The model compared it to a
|
|
|
|
|
|