Investigating Writing Professionals' Relationships with Generative AI: How Combined Perceptions of Rivalry and Collaboration Shape Work Practices and Outcomes

TL;DR

The paper investigates how writing professionals' dual perceptions of GenAI—as rivals and collaborators—shape their work practices and outcomes. Using a cross-sectional survey (n=403) plus RSA and PAM analyses, the authors show that rivalry and collaboration independently influence job crafting and skill maintenance, with collaboration more strongly linked to productivity and satisfaction. Importantly, a combined high rivalry and high collaboration profile yields the strongest, more balanced outcomes across crafting, skill maintenance, and productivity. The study advances a nuanced, bottom-up view of GenAI in professional writing and offers design implications—such as micro-frictions and communities of practice—to foster reflective, human-centered adaptation to GenAI. These insights have practical significance for shaping long-term career viability and responsible AI integration in creative professions.

Abstract

This study investigates how professional writers' complex relationship with GenAI shapes their work practices and outcomes. Through a cross-sectional survey with writing professionals (n=403) in diverse roles, we show that collaboration and rivalry orientation are associated with differences in work practices and outcomes. Rivalry is primarily associated with relational crafting and skill maintenance. Collaboration is primarily associated with task crafting, productivity, and satisfaction, at the cost of long-term skill deterioration. Combination of the orientations (high rivalry and high collaboration) reconciles these differences, while boosting the association with the outcomes. Our findings argue for a balanced approach where high levels of rivalry and collaboration are essential to shape work practices and generate outcomes aimed at the long-term success of the job. We present key design implications on how to increase friction (rivalry) and reduce over-reliance (collaboration) to achieve a more balanced relationship with GenAI.

Figures (6)

• Figure 1: The figure presents the distribution of the four profiles across professionals’ roles. The y-axis shows the four profiles, while the x-axis places roles along a spectrum ranging from those with a greater emphasis on creative aspects (left) to those with a greater emphasis on technical aspects (right). Proportions of the profiles in each role are reported at three levels: low, medium, and high.
• Figure 2: This figure presents the distributions of professionals' perceived use of GenAI in their work, categorized by the four profiles. Each panel presents the mean, confidence intervals, along with the response distributions for ten different question types where they use GenAI: 1) cognition; 2) creativity; 3) Relationships; 4) Core tasks; 5) Difficult tasks; 6) Time-pressured tasks; 7) Expert tasks; 8) Boring Tasks; 9) Non-essential tasks; 10) Tasks outside respondents' expertise. Participants responded on a scale of 1: Do it entirely by themselves to 7: Do it entirely with GenAI
• Figure 3: The figure presents visualization of the results from response surface analysis (RSA) for key aggregate measures, namely job crafting, skill maintenance, productivity, and satisfaction. The shades of blue indicate estimate ranges where the joint effects of rivalry and collaboration are the weakest, where as bright yellow indicates the strongest effects. Highest levels of collaboration and rivalry were required for job crafting, skill use, and productivity. Interestingly satisfaction and productivity also predicted by high levels of collaboration alone.
• Figure 4: The figure presents job crafting estimated means (EMMs) across profiles. Additionally Tukey's HSD test is used to show groups that are significantly different from each other. These are represented by the color and the letter. For all types of crafting except relational, profiles characterized by higher collaboration were associated with higher levels of crafting behavior. For relational crafting, only the HighR/HighC profile was associated with significantly higher crafting behavior.
• Figure 5: The figure presents skill maintenance estimated means (EMMs) across profiles. Additionally Tukey's HSD test is used to show groups that are significantly different from each other. These are represented by the color and the letter. For all types of skill maintenance except writing skill, only HighR/HighC correlated with substantial increase in skill maintenance.