Revision History
RLAIF · 8 revisions
Sizes are character counts of the article source. The signed number is the change from the previous revision.
Recent edit summaries
Detailed summaries recorded by editors. Generic maintenance summaries are omitted here; every recorded revision remains below. Dates describe the edit, not necessarily the event it covers.
- Correction: Corrected a fabricated Constitutional AI principle quote, misattributed findings (Singhal et al. length study described as RLAIF, a self-rewarding sycophancy claim not in the cited Sharma paper), the Llama 3 preference-data description, the start of Constitutional AI use in Claude (first Claude, not Claude 2), conflated same-size labeler results, and overstated RLAIF cost savings (the source estimates over 10x, not orders of magnitude) (plus 19 minor fixes)
Version 8 · Nov 5, 2026, 04:02 AM
- Improve internal links (2026-10-13)
Version 7 · Oct 13, 2026, 10:03 AM
- Added 6 contextual internal links
Version 6 · Jul 23, 2026, 03:31 PM
- Pre-update snapshot before enhancement to 4774 words.
Version 2 · May 17, 2026, 11:53 PM