ChatGPT for Zero-shot Dialogue State Tracking: A Solution or an Opportunity?

TL;DR

This paper investigates ChatGPT's potential in zero-shot dialogue state tracking, demonstrating state-of-the-art results but also discussing inherent limitations and future opportunities for specialized systems.

Contribution

It provides preliminary experimental evidence that ChatGPT achieves top zero-shot DST performance and discusses its limitations and future support roles for dedicated trackers.

Findings

ChatGPT achieves state-of-the-art zero-shot DST performance.

General purpose models have inherent limitations for replacing specialized systems.

In-context learning may support development of dynamic dialogue state trackers.

Abstract

Recent research on dialogue state tracking (DST) focuses on methods that allow few- and zero-shot transfer to new domains or schemas. However, performance gains heavily depend on aggressive data augmentation and fine-tuning of ever larger language model based architectures. In contrast, general purpose language models, trained on large amounts of diverse data, hold the promise of solving any kind of task without task-specific training. We present preliminary experimental results on the ChatGPT research preview, showing that ChatGPT achieves state-of-the-art performance in zero-shot DST. Despite our findings, we argue that properties inherent to general purpose models limit their ability to replace specialized systems. We further theorize that the in-context learning capabilities of such models will likely become powerful tools to support the development of dedicated and dynamic dialogue state trackers.