Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Cultural Perspectives and Expectations for Generative AI: A Global Survey Approach

Erin van Liemt, Renee Shelby, Andrew Smart, Sinchana Kumbale, Richard Zhang, Neha Dixit, Qazi Mamunur Rashid, Jamila Smith-Loud

TL;DR

This paper assesses understandings and beliefs about culture as it relates to GenAI from a large-scale global survey and concludes with a set of recommendations for Culture and GenAI development.

Abstract

There is a lack of empirical evidence about global attitudes around whether and how GenAI should represent cultures. This paper assesses understandings and beliefs about culture as it relates to GenAI from a large-scale global survey. We gathered data about what culture means to different groups, and about how GenAI should approach the representation of cultural artifacts, concepts, or values. We distill working definitions of culture directly from these communities to build an understanding of its conceptual complexities and how they relate to representations in Generative AI. We survey from across parts of Europe, North and South America, Asia, and Africa. We conclude with a set of recommendations for Culture and GenAI development. These include participatory approaches, prioritizing specific cultural dimensions beyond geography, such as religion and tradition, and a sensitivity framework for addressing cultural ``redlines''.

Cultural Perspectives and Expectations for Generative AI: A Global Survey Approach

TL;DR

This paper assesses understandings and beliefs about culture as it relates to GenAI from a large-scale global survey and concludes with a set of recommendations for Culture and GenAI development.

Abstract

There is a lack of empirical evidence about global attitudes around whether and how GenAI should represent cultures. This paper assesses understandings and beliefs about culture as it relates to GenAI from a large-scale global survey. We gathered data about what culture means to different groups, and about how GenAI should approach the representation of cultural artifacts, concepts, or values. We distill working definitions of culture directly from these communities to build an understanding of its conceptual complexities and how they relate to representations in Generative AI. We survey from across parts of Europe, North and South America, Asia, and Africa. We conclude with a set of recommendations for Culture and GenAI development. These include participatory approaches, prioritizing specific cultural dimensions beyond geography, such as religion and tradition, and a sensitivity framework for addressing cultural ``redlines''.
Paper Structure (30 sections, 5 figures, 6 tables)

This paper contains 30 sections, 5 figures, 6 tables.

Table of Contents

  1. Introduction
  2. Literature Review
  3. What is culture/what does it mean
  4. Technology and Culture
  5. Cultural Representation in Data
  6. Methodology
  7. Participants
  8. Study Design
  9. Survey Structure
  10. Survey Translation
  11. Data Analysis
  12. Quantitative Analysis
  13. Qualitative Analysis
  14. Results
  15. Understandings of Culture, Social Identities, & Important Aspects of Culture
  16. ...and 15 more sections

Figures (5)

  • Figure 1: Distribution of social identities selected as "important" by country shown as overall and individual percentages. A/B/C/D/E/F/G/H/I/J/K/L/M represent statistical significance at the 95% confidence level.
  • Figure 2: Social identities selected as important and sensitive for for all countries. This shows the relationship between the importance (Y-axis) and perceived GenAI sensitivity (X-axis). Darker shades indicate a high density of respondents who categorize a social identity as both culturally salient and technologically sensitive. Notably religion or tradition stands out as a salient identity
  • Figure 3: Social identities selected as important and sensitive for South Korea. This shows the relationship between the importance (Y-axis) and perceived GenAI sensitivity (X-axis). Darker shades indicate a high density of respondents who categorize a social identity as both culturally salient and technologically sensitive. Notably family stands out as a salient identity and health status is unique to this country.
  • Figure 4: Social identities selected as important and sensitive for UAE. This shows the relationship between the importance (Y-axis) and perceived GenAI sensitivity (X-axis). Darker shades indicate a high density of respondents who categorize a social identity as both culturally salient and technologically sensitive. Notably religion or tradition stands out as a salient identity and socioeconomic status is unique to this country.
  • Figure 5: Perception of GenAI harms by user familiarity. Distribution of responses identifying specific GenAI outputs as harmful, categorized by the participant's self-reported familiarity with the technology.