Beyond Demographics: Fine-tuning Large Language Models to Predict Individuals' Subjective Text Perceptions
This work addresses the problem of accurately modeling sociodemographic variation in subjective annotations for researchers and practitioners using LLMs, but it is incremental as it builds on prior findings of poor performance with sociodemographic prompting.
The study investigated whether large language models (LLMs) can be trained to predict individuals' subjective text perceptions based on sociodemographic attributes, finding that while performance improves with training, this is due to learning annotator-specific behavior rather than meaningful sociodemographic patterns, raising doubts about their use for simulating such variation.
People naturally vary in their annotations for subjective questions and some of this variation is thought to be due to the person's sociodemographic characteristics. LLMs have also been used to label data, but recent work has shown that models perform poorly when prompted with sociodemographic attributes, suggesting limited inherent sociodemographic knowledge. Here, we ask whether LLMs can be trained to be accurate sociodemographic models of annotator variation. Using a curated dataset of five tasks with standardized sociodemographics, we show that models do improve in sociodemographic prompting when trained but that this performance gain is largely due to models learning annotator-specific behaviour rather than sociodemographic patterns. Across all tasks, our results suggest that models learn little meaningful connection between sociodemographics and annotation, raising doubts about the current use of LLMs for simulating sociodemographic variation and behaviour.