Cultural Compass: A Framework for Organizing Societal Norms to Detect Violations in Human-AI Conversations

TL;DR

This paper introduces a detailed taxonomy and evaluation framework for assessing how well generative AI models adhere to sociocultural norms across different contexts and cultures, highlighting prevalent norm violations.

Contribution

It provides a comprehensive taxonomy of norms and an operational evaluation pipeline to measure AI adherence to sociocultural norms in realistic, open-ended interactions.

Findings

State-of-the-art models often violate norms

Violation rates vary by model, context, and country

Evaluation framework enables nuanced norm assessment

Abstract

Generative AI models ought to be useful and safe across cross-cultural contexts. One critical step toward this goal is understanding how AI models adhere to sociocultural norms. While this challenge has gained attention in NLP, existing work lacks both nuance and coverage in understanding and evaluating models' norm adherence. We address these gaps by introducing a taxonomy of norms that clarifies their contexts (e.g., distinguishing between human-human norms that models should recognize and human-AI interactional norms that apply to the human-AI interaction itself), specifications (e.g., relevant domains), and mechanisms (e.g., modes of enforcement). We demonstrate how our taxonomy can be operationalized to automatically evaluate models' norm adherence in naturalistic, open-ended settings. Our exploratory analyses suggest that state-of-the-art models frequently violate norms, though violation rates vary by model, interactional context, and country. We further show that violation rates also vary by prompt intent and situational framing. Our taxonomy and demonstrative evaluation pipeline enable nuanced, context-sensitive evaluation of cultural norm adherence in realistic settings.