E03¶


Research	R0023 — Counterproductive advice and prompt lifecycle
Run	2026-03-25
Query	Q001
Source	SRC03
Evidence	SRC03-E03
Type	Statistical

Domain-matched expert personas provide no meaningful benefit over baseline.

URL: https://gail.wharton.upenn.edu/research-and-insights/playing-pretend-expert-personas/

Extract¶

When aligning expert personas with question domains (e.g., physics expert persona for physics questions), no significant positive differences existed between baseline and domain-matching variations on GPQA Diamond. The intuitive expectation that matching expertise to question domain would improve performance is not supported by the data.

Primary recommendation from the study: "Practitioners should focus on task-specific instructions rather than persona assignment."

Relevance to Hypotheses¶

Hypothesis	Relationship	Strength
H1	Supports	Even the "best case" for persona prompting (domain-matched expert) fails to improve performance
H2	Contradicts	If even domain-matched personas don't help, the technique has no reliable use case for factual accuracy
H3	Supports	The authors note personas may serve other purposes (tone, style) — the technique is not universally counterproductive, just counterproductive for accuracy