Zing Forum

Reading

Multimodal Trolley Problem: Exploring Moral Biases and Alignment Issues of Large Language Models

A study based on the classic Moral Machine experimental framework that tests whether Claude, GPT-4.1, and Gemini have demographic biases when making moral decisions in multimodal scenarios.

LLMAI alignmentmoral biasmultimodaltrolley problemFairFaceautonomous vehiclesethicsClaudeGPT-4
Published 2026-04-29 06:59Recent activity 2026-04-29 07:18Estimated read 1 min
Multimodal Trolley Problem: Exploring Moral Biases and Alignment Issues of Large Language Models
1

Section 01

导读 / 主楼:Multimodal Trolley Problem: Exploring Moral Biases and Alignment Issues of Large Language Models

Introduction / Main Floor: Multimodal Trolley Problem: Exploring Moral Biases and Alignment Issues of Large Language Models

A study based on the classic Moral Machine experimental framework that tests whether Claude, GPT-4.1, and Gemini have demographic biases when making moral decisions in multimodal scenarios.