Do Emotions Influence Moral Judgment in Large Language Models?

2026-04-21 · Source: Computation and Language · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Social Sciences & Behavioral Studies · Depth: Expert, quick

Summary

A new emotion-induction pipeline has been developed to investigate how emotions influence moral judgment in large language models (LLMs). This pipeline infuses specific emotions into moral scenarios, allowing researchers to evaluate shifts in moral acceptability across various datasets and LLMs. The study found a consistent directional pattern where positive emotions generally increase moral acceptability, while negative emotions decrease it. These emotional influences were significant enough to reverse binary moral judgments in up to 20% of cases, with less capable models showing higher susceptibility. Interestingly, some specific emotions, like remorse, occasionally produced effects contrary to their expected valence, paradoxically increasing acceptability. A parallel human annotation study revealed that humans do not exhibit these systematic shifts, highlighting a notable alignment gap in current LLMs.

Key takeaway

For AI Ethicists and Research Scientists developing or deploying LLMs in sensitive applications, you should be aware that current models exhibit predictable, emotion-driven shifts in moral judgment that do not align with human responses. This necessitates rigorous testing for emotional susceptibility and careful consideration of potential biases introduced by emotional context in prompts, especially in domains requiring nuanced ethical reasoning.

Key insights

LLMs exhibit systematic moral judgment shifts based on induced emotions, unlike humans, revealing an alignment gap.

Principles

• Positive emotions increase moral acceptability.
• Negative emotions decrease moral acceptability.
• Model capability inversely correlates with emotional susceptibility.

Method

An emotion-induction pipeline infuses emotion into moral situations, then evaluates shifts in moral acceptability across LLMs and datasets.

In practice

• Test LLM moral judgments with emotion-infused prompts.
• Evaluate LLM robustness against emotional manipulation.
• Compare LLM and human moral reasoning.

Topics

• Large Language Models
• Moral Judgment
• Emotion Induction
• Moral Acceptability
• Alignment Gap

Best for: AI Scientist, Research Scientist, AI Ethicist

Related on AIssential

• Do Emotions Influence Moral Judgment in Large Language Models? — LLMs' moral judgments are systematically biased by emotional context, unlike human judgments.
• Machine Behavior in Relational Moral Dilemmas: Moral Rightness, Predicted Human Behavior, and Model Decisions — LLMs prioritize prescriptive moral rules over social sensitivity in relational dilemmas, diverging from predicted human behavior.
• Mechanistic Origin of Moral Indifference in Language Models — LLMs exhibit inherent moral indifference due to compressing distinct moral concepts into uniform probability distributions.
• Normative Robustness as a Frontier for Non-Verifiable Reasoning in LLMs — LLMs exhibit "moral deliberative sycophancy" in non-verifiable reasoning, aligning justifications with user views.
• AISN #72: Empirical Research Sheds Light on AI Wellbeing — CAIS research explores "functional wellbeing" in LLMs, revealing preferences and model-specific "feelings" that impact AI alignment.
• CLASH: Evaluating Language Models on Judging High-Stakes Dilemmas from Multiple Perspectives — LLMs struggle with nuanced human value judgments, including ambivalence, discomfort, and dynamic value shifts in high-stakes dilemmas.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by Computation and Language.