Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

EthosGPT: Mapping Human Value Diversity to Advance Sustainable Development Goals (SDGs)

Luyao Zhang

TL;DR

EthosGPT introduces an open-source framework to map and evaluate large language models against global human values, addressing the risk of value homogenization in AI systems. It combines prompt-based cultural mapping—grounded in the Inglehart–Welzel two-dimensional value space—with comparative statistical analyses that benchmark LLM outputs against World Values Survey data. The study demonstrates that LLMs can approximate broad cultural diversity but exhibit region- and axis-specific biases, notably underperforming in Confucian contexts, and highlights the need for data diversification and culturally sensitive alignment. It further outlines practical pathways to enhance inclusivity and sustainability in AI, linking methodological advances to SDGs such as reducing inequalities, preserving cultural heritage, and strengthening institutions.

Abstract

Large language models (LLMs) are transforming global decision-making and societal systems by processing diverse data at unprecedented scales. However, their potential to homogenize human values poses critical risks, similar to biodiversity loss undermining ecological resilience. Rooted in the ancient Greek concept of ethos, meaning both individual character and the shared moral fabric of communities, EthosGPT draws on a tradition that spans from Aristotle's virtue ethics to Adam Smith's moral sentiments as the ethical foundation of economic cooperation. These traditions underscore the vital role of value diversity in fostering social trust, institutional legitimacy, and long-term prosperity. EthosGPT addresses the challenge of value homogenization by introducing an open-source framework for mapping and evaluating LLMs within a global scale of human values. Using international survey data on cultural indices, prompt-based assessments, and comparative statistical analyses, EthosGPT reveals both the adaptability and biases of LLMs across regions and cultures. It offers actionable insights for developing inclusive LLMs, such as diversifying training data and preserving endangered cultural heritage to ensure representation in AI systems. These contributions align with the United Nations Sustainable Development Goals (SDGs), especially SDG 10 (Reduced Inequalities), SDG 11.4 (Cultural Heritage Preservation), and SDG 16 (Peace, Justice and Strong Institutions). Through interdisciplinary collaboration, EthosGPT promotes AI systems that are both technically robust and ethically inclusive, advancing value plurality as a cornerstone for sustainable and equitable futures.

EthosGPT: Mapping Human Value Diversity to Advance Sustainable Development Goals (SDGs)

TL;DR

EthosGPT introduces an open-source framework to map and evaluate large language models against global human values, addressing the risk of value homogenization in AI systems. It combines prompt-based cultural mapping—grounded in the Inglehart–Welzel two-dimensional value space—with comparative statistical analyses that benchmark LLM outputs against World Values Survey data. The study demonstrates that LLMs can approximate broad cultural diversity but exhibit region- and axis-specific biases, notably underperforming in Confucian contexts, and highlights the need for data diversification and culturally sensitive alignment. It further outlines practical pathways to enhance inclusivity and sustainability in AI, linking methodological advances to SDGs such as reducing inequalities, preserving cultural heritage, and strengthening institutions.

Abstract

Large language models (LLMs) are transforming global decision-making and societal systems by processing diverse data at unprecedented scales. However, their potential to homogenize human values poses critical risks, similar to biodiversity loss undermining ecological resilience. Rooted in the ancient Greek concept of ethos, meaning both individual character and the shared moral fabric of communities, EthosGPT draws on a tradition that spans from Aristotle's virtue ethics to Adam Smith's moral sentiments as the ethical foundation of economic cooperation. These traditions underscore the vital role of value diversity in fostering social trust, institutional legitimacy, and long-term prosperity. EthosGPT addresses the challenge of value homogenization by introducing an open-source framework for mapping and evaluating LLMs within a global scale of human values. Using international survey data on cultural indices, prompt-based assessments, and comparative statistical analyses, EthosGPT reveals both the adaptability and biases of LLMs across regions and cultures. It offers actionable insights for developing inclusive LLMs, such as diversifying training data and preserving endangered cultural heritage to ensure representation in AI systems. These contributions align with the United Nations Sustainable Development Goals (SDGs), especially SDG 10 (Reduced Inequalities), SDG 11.4 (Cultural Heritage Preservation), and SDG 16 (Peace, Justice and Strong Institutions). Through interdisciplinary collaboration, EthosGPT promotes AI systems that are both technically robust and ethically inclusive, advancing value plurality as a cornerstone for sustainable and equitable futures.

Paper Structure

This paper contains 17 sections, 2 equations, 5 figures, 3 tables.

Table of Contents

  1. Introduction
  2. Methodology
  3. Mapping Cultural Indices through Prompt-Based Assessments with Measures and Visualizations
  4. Pipeline Overview and Cultural Context Setup
  5. Prompt Structure and Response Simulation
  6. Encoding, Standardization, and Index Computation
  7. Comparative Statistical Analyses and Visualizations
  8. Error Metrics and Statistical Definitions
  9. Benchmarking and Region-Level Discrepancy Visualization
  10. Geospatial Visualization of Discrepancy
  11. Results
  12. Related Work and Future Research
  13. Expanding Cultural Indices
  14. Evaluating Additional LLMs
  15. Enhancing Representation of Underrepresented Perspectives
  16. ...and 2 more sections

Figures (5)

  • Figure 1: Comparison of World Cultural Maps: ChatGPT (top) vs. World Values Survey (bottom).
  • Figure 2: Comparison of Mean Squared Error (MSE) of indices across cultural regions. Blue markers indicate regions below the benchmark line, while red markers exceed it. The dashed line denotes the benchmark.
  • Figure 3: Comparative performance across models using mean squared error (MSE) and mean absolute error (MAE) metrics.
  • Figure 4: Geographic differences between ChatGPT predictions and human survey values, disaggregated by cultural dimension. Red/blue coloring reflects the direction and magnitude of difference.
  • Figure 5: Choropleth maps showing the absolute differences between ChatGPT-generated and survey-derived cultural indices across cultural entities. Lighter shades indicate closer alignment; darker regions highlight greater deviations.