Seeing Through the Fog: A Cost-Effectiveness Analysis of Hallucination Detection Systems

TL;DR

This paper compares various hallucination detection systems for large language models, analyzing their effectiveness and cost, and emphasizes the need for scalable solutions that balance performance and resource use.

Contribution

It introduces a comprehensive evaluation framework for hallucination detection systems using DOR and cost-effectiveness metrics, highlighting the importance of adaptable solutions across model sizes.

Findings

Advanced detection models perform better but are more costly

Performance must be maintained across different model sizes

Cost-effectiveness varies significantly among detection systems

Abstract

This paper presents a comparative analysis of hallucination detection systems for AI, focusing on automatic summarization and question answering tasks for Large Language Models (LLMs). We evaluate different hallucination detection systems using the diagnostic odds ratio (DOR) and cost-effectiveness metrics. Our results indicate that although advanced models can perform better they come at a much higher cost. We also demonstrate how an ideal hallucination detection system needs to maintain performance across different model sizes. Our findings highlight the importance of choosing a detection system aligned with specific application needs and resource constraints. Future research will explore hybrid systems and automated identification of underperforming components to enhance AI reliability and efficiency in detecting and mitigating hallucinations.