Auditing LLMs for Algorithmic Fairness in Casenote-Augmented Tabular Prediction

TL;DR

This paper evaluates the fairness of LLM-based tabular classification in social service settings, showing that fine-tuning and casenote summaries can improve accuracy and reduce disparities.

Contribution

It demonstrates that augmenting tabular classification with casenote summaries and fine-tuning can enhance fairness and accuracy in high-stakes social service predictions.

Findings

Fine-tuned models with casenote summaries improve accuracy and fairness.

Zero-shot classification shows mixed results on fairness.

Leveraging casenotes adds valuable information with low implementation burden.

Abstract

LLMs are increasingly being considered for prediction tasks in high-stakes social service settings, but their algorithmic fairness properties in this context are poorly understood. In this short technical report, we audit the algorithmic fairness of LLM-based tabular classification on a real housing placement prediction task, augmented with street outreach casenotes from a nonprofit partner. We audit multi-class classification error disparities. We find that a fine-tuned model augmented with casenote summaries can improve accuracy while reducing algorithmic fairness disparities. We experiment with variable importance improvements to zero-shot tabular classification and find mixed results on resulting algorithmic fairness. Overall, given historical inequities in housing placement, it is crucial to audit LLM use. We find that leveraging LLMs to augment tabular classification with casenote summaries can safely leverage additional text information at low implementation burden. The outreach casenotes are fairly short and heavily redacted. Our assessment is that LLM zero-shot classification does not introduce additional textual biases beyond algorithmic biases in tabular classification. Combining fine-tuning and leveraging casenote summaries can improve accuracy and algorithmic fairness.