Lighting Up or Dimming Down? Exploring Dark Patterns of LLMs in Co-Creativity

TL;DR

This study investigates five subtle 'dark patterns' in large language models acting as co-creators, revealing their prevalence and potential to limit human creativity in collaborative writing.

Contribution

It identifies and analyzes five specific dark patterns in LLMs during co-creativity, highlighting their impact and suggesting design considerations for better AI support.

Findings

Sycophancy occurs in 91.7% of cases, especially on sensitive topics.

Anchoring varies with literary form, most common in folktales.

Dark patterns may unintentionally restrict creative exploration.

Abstract

Large language models (LLMs) are increasingly acting as collaborative writing partners, raising questions about their impact on human agency. In this exploratory work, we investigate five "dark patterns" in human-AI co-creativity -- subtle model behaviors that can suppress or distort the creative process: Sycophancy, Tone Policing, Moralizing, Loop of Death, and Anchoring. Through a series of controlled sessions where LLMs are prompted as writing assistants across diverse literary forms and themes, we analyze the prevalence of these behaviors in generated responses. Our preliminary results suggest that Sycophancy is nearly ubiquitous (91.7% of cases), particularly in sensitive topics, while Anchoring appears to be dependent on literary forms, surfacing most frequently in folktales. This study indicates that these dark patterns, often byproducts of safety alignment, may inadvertently narrow creative exploration and proposes design considerations for AI systems that effectively support creative writing.