Scenario-based dilemmas combined with activation steering probe and shift LLM values along Inglehart-Welzel axes, revealing persistent entanglement between dimensions that mirrors human survey data.
Griffiths, and Arvind Narayanan
5 Pith papers cite this work. Polarity classification is still indexing.
5
Pith papers citing it
citation-role summary
background 1
citation-polarity summary
years
2026 5verdicts
UNVERDICTED 5roles
background 1polarities
background 1representative citing papers
C-3PO preference optimisation raises Singleton Fleiss κ_S by up to 0.13, cutting cross-lingual cultural inconsistency in multilingual LLMs without erasing persona or cultural knowledge.
HiMed releases a Hindi medical reasoning corpus and benchmark and shows that training an 8B LLM with decaying scaffolding reward improves Hindi performance and narrows the English-Hindi accuracy gap.
Generative AI should be evaluated through computational hermeneutics using iterative, human-inclusive benchmarks that measure cultural context rather than isolated model outputs.
Proposes AI-driven simulations for literary-historical experiments and reports preliminary text-generation results claiming the first limited in-distribution outputs matching human novels.
citing papers explorer
-
Scenario-based Probing and Steering Cultural Values in Large Language Models--Extended Version
Scenario-based dilemmas combined with activation steering probe and shift LLM values along Inglehart-Welzel axes, revealing persistent entanglement between dimensions that mirrors human survey data.
-
Mitigating Cross-Lingual Cultural Inconsistencies in LLMs via Consensus-Driven Preference Optimisation
C-3PO preference optimisation raises Singleton Fleiss κ_S by up to 0.13, cutting cross-lingual cultural inconsistency in multilingual LLMs without erasing persona or cultural knowledge.
-
HiMed: Incentivizing Hindi Reasoning in Medical LLMs
HiMed releases a Hindi medical reasoning corpus and benchmark and shows that training an 8B LLM with decaying scaffolding reward improves Hindi performance and narrows the English-Hindi accuracy gap.
-
Computational Hermeneutics: Evaluating generative AI as a cultural technology
Generative AI should be evaluated through computational hermeneutics using iterative, human-inclusive benchmarks that measure cultural context rather than isolated model outputs.
-
AI as a Tool for Simulation-Based Experiments in Literary Studies
Proposes AI-driven simulations for literary-historical experiments and reports preliminary text-generation results claiming the first limited in-distribution outputs matching human novels.