REVIEW 6 cited by
Unintended Impacts of LLM Alignment on Global Representation
Not yet reviewed by Pith; the record is open.
This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.
SPECIMEN: schema-true, not a live event
T0 review · schema-true
One-sentence machine reading of the paper's core claim.
pith:XXXXXXXX · record.json · timestamp
Unintended Impacts of LLM Alignment on Global Representation
read the original abstract
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.
Forward citations
Cited by 6 Pith papers
-
CCBENCH: Assessing LLM Cultural Competence via Implicitly Signaled Norms using Health Queries
Leading LLMs produce culturally appropriate health responses only 20-30% of the time on a new continuum-of-norm-adherence benchmark, with a strong bias toward Western defaults.
-
AI-Augmented Surveys: Leveraging Large Language Models and Surveys for Opinion Prediction
LLM embeddings enable strong retrodiction of masked GSS opinions via cross-validation and external validation but only modest performance on entirely unasked opinions.
-
Probing Latent Colombian Identity Inferences in Qwen2.5-7B with Natural Language Autoencoders
A pilot probe finds weak, non-robust evidence that Qwen2.5-7B internally represents Colombian identity from a single implicit cue; the only nominally significant effect is driven by unrestricted, confabulated national...
-
Representational Harms in LLM-Generated Narratives Against Global Majority Nationalities
LLMs generate narratives containing persistent stereotypes, erasure, and one-dimensional portrayals of Global Majority national identities, with minoritized groups overrepresented in subordinated roles by more than fi...
-
Attributing Culture-Conditioned Generations to Pretraining Corpora
MEMOed framework attributes LLM generations about cultures to pretraining memorization and finds frequency-based biases across 110 cultures for food and clothing.
-
SemEval-2026 Task 7: Everyday Knowledge Across Diverse Languages and Cultures
SemEval-2026 Task 7 presents a benchmark and two evaluation tracks for assessing LLMs on everyday knowledge in diverse languages and cultures without allowing training on the test data.
discussion (0)
Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.