E01¶


Research	R0057 — RLHF Yes-Men Claims v3
Run	2026-04-01
Claim	C022
Source	SRC01
Evidence	SRC01-E01
Type	Analytical

Vocabulary gap exists; some bridging attempts but no widely adopted shared vocabulary

URL: https://cset.georgetown.edu/publication/ai-safety-and-automation-bias/

Extract¶

Georgetown CSET published an issue brief connecting AI safety and automation bias. A 2026 medRxiv paper introduces 'structural drift' as a bridging concept. However, neither has achieved widespread adoption as a shared vocabulary across AI safety and human factors communities.

Relevance to Hypotheses¶

Hypothesis	Relationship	Strength
H1	Supports	Directly addresses claim accuracy
H2	Supports	Allows for partial correctness
H3	Contradicts	Evidence contradicts material inaccuracy

Context¶

Some bridging work exists but the claim that 'no shared vocabulary bridges them' is slightly overstated.