Unintended Impacts of LLM Alignment on Global Representation
This addresses unintended representation biases in LLM alignment for developers and users, though it is incremental as it builds on existing alignment methods.
The study investigated how alignment procedures like RLHF and DPO affect LLM performance across English dialects, multilingualism, and global opinions, finding they create disparities between English dialects and global opinions while improving capabilities in several languages.
Before being deployed for user-facing applications, developers align Large Language Models (LLMs) to user preferences through a variety of procedures, such as Reinforcement Learning From Human Feedback (RLHF) and Direct Preference Optimization (DPO). Current evaluations of these procedures focus on benchmarks of instruction following, reasoning, and truthfulness. However, human preferences are not universal, and aligning to specific preference sets may have unintended effects. We explore how alignment impacts performance along three axes of global representation: English dialects, multilingualism, and opinions from and about countries worldwide. Our results show that current alignment procedures create disparities between English dialects and global opinions. We find alignment improves capabilities in several languages. We conclude by discussing design decisions that led to these unintended impacts and recommendations for more equitable preference tuning. We make our code and data publicly available on Github.