Shaping Human-AI Collaboration: Varied Scaffolding Levels in Co-writing with Language Models

TL;DR

This study investigates how different levels of AI assistance in co-writing affect writing quality, productivity, and user experience, revealing that high scaffolding improves outcomes especially for non-regular writers, with implications for personalized AI tool design.

Contribution

The paper introduces a comprehensive experimental analysis of varied AI scaffolding levels in co-writing, highlighting the benefits of high scaffolding and the importance of personalization.

Findings

High scaffolding improves writing quality and productivity for non-regular writers.

Low scaffolding does not significantly impact writing outcomes.

No significant cognitive burden observed, but some decrease in text ownership and satisfaction.

Abstract

Advances in language modeling have paved the way for novel human-AI co-writing experiences. This paper explores how varying levels of scaffolding from large language models (LLMs) shape the co-writing process. Employing a within-subjects field experiment with a Latin square design, we asked participants (N=131) to respond to argumentative writing prompts under three randomly sequenced conditions: no AI assistance (control), next-sentence suggestions (low scaffolding), and next-paragraph suggestions (high scaffolding). Our findings reveal a U-shaped impact of scaffolding on writing quality and productivity (words/time). While low scaffolding did not significantly improve writing quality or productivity, high scaffolding led to significant improvements, especially benefiting non-regular writers and less tech-savvy users. No significant cognitive burden was observed while using the scaffolded writing tools, but a moderate decrease in text ownership and satisfaction was noted. Our results have broad implications for the design of AI-powered writing tools, including the need for personalized scaffolding mechanisms.