Skip to content

Era-aware AI vocabulary breakdown + speculative gap-filling pattern#111

Open
philippdubach wants to merge 1 commit intoblader:mainfrom
philippdubach:era-vocab-and-gap-filling
Open

Era-aware AI vocabulary breakdown + speculative gap-filling pattern#111
philippdubach wants to merge 1 commit intoblader:mainfrom
philippdubach:era-vocab-and-gap-filling

Conversation

@philippdubach
Copy link
Copy Markdown

Summary

Two narrowly scoped updates sourced from the current revision of Wikipedia: Signs of AI writing (revision fetched 2026-05-01).

  • §7 (AI Vocabulary): Replaces the flat high-frequency word list with the era-specific clusters now documented on the wiki page (GPT-4 / GPT-4o / GPT-5 eras). Adds bolstered and meticulous/meticulously to the master list, plus a one-line caveat about literal vs figurative usage (e.g., underscore as a literal underline, delve in geology).
  • §21 (renamed to "Knowledge-Cutoff Disclaimers and Speculative Gap-Filling"): Covers the newer retrieval-augmented pattern where a model, having failed to find a source, writes a paragraph about not having found one and then speculates that the subject "maintains a low profile" or "keeps personal details private." Adds a second before/after example for the gap-filling case.
  • README: Tightens the §21 row label to reflect both subpatterns.

No new patterns; pattern count stays at 29. No version bump — happy to defer that to whatever coordination you do with the open v2.6.0 PRs (#85, #98).

Test plan

  • Diff is two files; SKILL.md and README.md
  • §7 keeps its existing Before/After example unchanged
  • §21 keeps its existing Before/After example as the cutoff-disclaimer case, and adds a separate gap-filling Before/After
  • Pattern numbering and section anchors are unchanged
  • Skill loads in Claude Code with no parse errors

Source: Wikipedia:Signs of AI writing — see "High density of AI vocabulary words" and "Knowledge-cutoff disclaimers and speculation about gaps in sources" sections.

…tern

Two changes sourced from Wikipedia: Signs of AI writing (revision
fetched 2026-05-01).

§7 (AI Vocabulary): replace the flat high-frequency word list with
the era-specific clusters now documented on the wiki page (GPT-4 /
GPT-4o / GPT-5 eras). Add 'bolstered' and 'meticulous/meticulously'
to the master list, and a one-line caveat about literal vs
figurative usage.

§21 (renamed to "Knowledge-Cutoff Disclaimers and Speculative
Gap-Filling"): cover the newer retrieval-augmented pattern where
the model, having failed to find a source, writes a paragraph about
not having found one and then speculates that the subject
"maintains a low profile" or "keeps personal details private."
Adds a second before/after example for the gap-filling case.

README: tighten the §21 row label to reflect both subpatterns.

No version bump (leaving that to the maintainer to coordinate with
the open v2.6.0 PRs). No new patterns; pattern count stays at 29.
philippdubach added a commit to philippdubach/humanizer that referenced this pull request May 1, 2026
Brings the fork's main branch in line with the maintained local
v2.6.0, consolidating the changes that are also opened as focused
PRs against blader/humanizer (blader#111, blader#112, blader#113):

- §7 expanded with era-specific AI vocabulary clusters (GPT-4 /
  GPT-4o / GPT-5 eras), plus 'bolstered' and 'meticulous' added to
  the master list and a literal-vs-figurative caveat.
- §21 renamed to "Knowledge-Cutoff Disclaimers and Speculative
  Gap-Filling"; covers the retrieval-augmented "maintains a low
  profile" / "keeps personal details private" speculation pattern.
- New patterns §30-34: reference-markup artifacts (turn0search0,
  oaicite, utm_source=chatgpt.com, etc.), placeholder leftovers,
  Markdown/wikitext contamination, formal "Conclusion" closers,
  didactic disclaimers.
- New Detection Guidance group: what NOT to flag (false positives),
  signs of human writing to preserve, and per-model LLM idiolects.

Frontmatter version bumped to 2.6.0. README pattern table updated
(29 → 34 patterns) with a new Artifacts and Contamination section
and a pointer to Detection Guidance. WARP.md count corrected from
the stale "25 patterns" to 34.

Sourced from Wikipedia: Signs of AI writing (revision fetched
2026-05-01).
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant