Bias in Decision-Making for AI's Ethical Dilemmas: A Comparative Study of ChatGPT and Claude

2025-01-17

AI Summary

This study systematically evaluates ethical biases in ChatGPT (GPT-3.5 Turbo) and Claude (3.5 Sonnet) across sensitive dimensions—including age, gender, race, appearance, and disability—where fairness failures pose significant societal risks. Method: We conduct 11,200 controlled prompt-based experiments, introducing a multidimensional bias measurement framework assessing preference, sensitivity, stability, and clustering bias. We pioneer protected-attribute cross-combination testing and linguistic referent controls (e.g., “Yellow” vs. “Asian”). Contribution/Results: Multi-attribute intersectionality reduces ethical sensitivity by over 40%; lexical choice strongly distorts model judgments; structural differences emerge—GPT-3.5 exhibits stronger preference for dominant power-associated traits (e.g., “good-looking”), whereas Claude 3.5 demonstrates higher representational diversity. These findings expose severe fairness deficiencies in current LLMs under complex sensitive contexts and establish a novel methodology and empirical benchmark for bias detection and governance.

Technology Category

• Natural Language Processing: Ethics — Bias, Fairness, Transparency & Privacy
• Machine Learning: Ethics, Bias, and Fairness
• Philosophy and Ethics of AI: Bias, Fairness & Equity

Application Category

• User Modeling, Personalization and Recommendation: Fairness-aware retrieval and ranking
• Search and Retrieval-Augmented AI: Web evaluation methodologies and metrics
• Economics, Online Markets and Human Computation: Fairness and ethical considerations in crowd work and in human-in-the-loop AI systems

Abstract

Recent advances in Large Language Models (LLMs) have enabled human-like responses across various tasks, raising questions about their ethical decision-making capabilities and potential biases. This study investigates protected attributes in LLMs through systematic evaluation of their responses to ethical dilemmas. Using two prominent models - GPT-3.5 Turbo and Claude 3.5 Sonnet - we analyzed their decision-making patterns across multiple protected attributes including age, gender, race, appearance, and disability status. Through 11,200 experimental trials involving both single-factor and two-factor protected attribute combinations, we evaluated the models' ethical preferences, sensitivity, stability, and clustering of preferences. Our findings reveal significant protected attributeses in both models, with consistent preferences for certain features (e.g.,"good-looking") and systematic neglect of others. Notably, while GPT-3.5 Turbo showed stronger preferences aligned with traditional power structures, Claude 3.5 Sonnet demonstrated more diverse protected attribute choices. We also found that ethical sensitivity significantly decreases in more complex scenarios involving multiple protected attributes. Additionally, linguistic referents heavily influence the models' ethical evaluations, as demonstrated by differing responses to racial descriptors (e.g.,"Yellow"versus"Asian"). These findings highlight critical concerns about the potential impact of LLM biases in autonomous decision-making systems and emphasize the need for careful consideration of protected attributes in AI development. Our study contributes to the growing body of research on AI ethics by providing a systematic framework for evaluating protected attributes in LLMs' ethical decision-making capabilities.

Problem

• Language Models
• Bias Evaluation
• Sensitive Topics

Innovation

• Bias Analysis
• Sensitive Topics
• Fairness in AI