A small lab study found GPT-4o-mini debates using a combined personalization-plus-fabricated-stats strategy shifted opinions in 51% of trials versus 32% for static human arguments, an effect not clearly supported by significance tests.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CL 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Tailored Truths: Optimizing LLM Persuasion with Personalization and Fabricated Statistics
A small lab study found GPT-4o-mini debates using a combined personalization-plus-fabricated-stats strategy shifted opinions in 51% of trials versus 32% for static human arguments, an effect not clearly supported by significance tests.