Help Me Write a Story: Evaluating LLMs' Ability to Generate Writing Feedback

TL;DR

This paper assesses how well large language models can provide meaningful writing feedback to creative writers, highlighting their strengths and limitations through a new dataset and evaluation framework.

Contribution

It introduces a novel dataset, task definition, and evaluation methods for assessing LLMs' ability to give writing feedback on corrupted stories.

Findings

Models provide mostly accurate and specific feedback

Models often miss the main writing issues

Models struggle to decide when to give critical versus positive feedback

Abstract

Can LLMs provide support to creative writers by giving meaningful writing feedback? In this paper, we explore the challenges and limitations of model-generated writing feedback by defining a new task, dataset, and evaluation frameworks. To study model performance in a controlled manner, we present a novel test set of 1,300 stories that we corrupted to intentionally introduce writing issues. We study the performance of commonly used LLMs in this task with both automatic and human evaluation metrics. Our analysis shows that current models have strong out-of-the-box behavior in many respects -- providing specific and mostly accurate writing feedback. However, models often fail to identify the biggest writing issue in the story and to correctly decide when to offer critical vs. positive feedback.