On Second Thought, Let's Not Think Step by Step! Bias and Toxicity in Zero-Shot Reasoning

TL;DR

This paper investigates how zero-shot Chain of Thought prompting affects large language models' behavior in socially sensitive contexts, revealing increased risks of harmful outputs and biases, especially with larger models.

Contribution

It provides a controlled evaluation of zero-shot CoT in socially sensitive domains, highlighting potential risks and the influence of model size and instruction following.

Findings

Zero-shot CoT increases harmful outputs in sensitive domains.

Larger models tend to produce more harmful CoTs.

Better instruction following reduces harmful biases.

Abstract

Generating a Chain of Thought (CoT) has been shown to consistently improve large language model (LLM) performance on a wide range of NLP tasks. However, prior work has mainly focused on logical reasoning tasks (e.g. arithmetic, commonsense QA); it remains unclear whether improvements hold for more diverse types of reasoning, especially in socially situated contexts. Concretely, we perform a controlled evaluation of zero-shot CoT across two socially sensitive domains: harmful questions and stereotype benchmarks. We find that zero-shot CoT reasoning in sensitive domains significantly increases a model's likelihood to produce harmful or undesirable output, with trends holding across different prompt formats and model variants. Furthermore, we show that harmful CoTs increase with model size, but decrease with improved instruction following. Our work suggests that zero-shot CoT should be used with caution on socially important tasks, especially when marginalized groups or sensitive topics are involved.