Co-Designing Collaborative Generative AI Tools for Freelancers

Paper Title

Co-Designing Collaborative Generative AI Tools for Freelancers

Publication Info

• Topic area: Generative AI tools for freelancer collaboration
• Keywords: Generative AI, freelancer collaboration, co-design, auxiliary AI, technological rationality, creative agency, DALL-E, participatory design, critical theory, productive friction

Background and Problem

• Problem / challenge: Current generative AI tools prioritize individual productivity and efficiency, offering limited support for the decentralized, temporary, and trust-based collaborations typical of freelancers. These tools risk undermining freelancers' creative agency, work identities, and collaborative practices.
• Significance: Freelancers represent a significant portion of the global workforce (over 1.5 billion people) and rely on collaboration for tasks such as joint bidding, creative co-production, and peer learning. Effective AI tools could enhance these practices, but poorly designed systems may exacerbate isolation and precarity.
• Motivation and related work: While existing research and tools like ChatGPT Teams and GitHub Copilot Workspace focus on structured organizational teams, they fail to address freelancers' unique needs. Prior studies highlight the risks of over-reliance on AI, lack of contextual understanding, and ethical concerns, but little work explores how generative AI can be tailored to freelancers' decentralized collaborations.

Solution

• Proposed approach: The study proposes "auxiliary AI" systems that support freelancers' creative agency and collaboration without dominating decision-making. These systems emphasize productive friction, contextual understanding, and human-led workflows.
• Novelty:
  1. Empirical insights into how current generative AI tools fail to support freelancer collaboration.
  2. Identification of ethical challenges, including risks to creative agency and originality.
  3. Design principles for collaborative AI tools, emphasizing auxiliary roles, diverse skill support, and human coordination.
• Procedure and key techniques:
  • Conducted co-design sessions with 27 freelancers using the Future Workshops methodology.
  • Used DALL-E as a design probe to help participants visualize and iterate on their ideas for collaborative AI tools.
  • Combined synchronous Zoom sessions and asynchronous Slack discussions to refine design concepts over several months.

Results

• Concrete findings:
  • Current generative AI tools fail to understand freelancers' collaborative contexts, producing generic outputs that lack relevance.
  • Over-reliance on AI can undermine creativity, critical thinking, and decision-making in collaborative projects.
  • AI-generated content risks plagiarism, loss of creative agency, and diminished originality in joint work.
• Advantage over baselines:
  • Proposed auxiliary AI systems address the limitations of existing tools by supporting human-led collaboration, preserving creative authenticity, and fostering productive friction.
• Experiments / evaluation:
  • Co-design sessions involved 27 freelancers from diverse fields, using DALL-E to generate and refine visual probes.
  • Participants critiqued current AI tools, envisioned future systems, and developed actionable design principles.
• Limitations and future work:
  • The study did not include freelancers strongly skeptical of AI, limiting the diversity of perspectives.
  • Future work should explore longitudinal impacts of AI tools on freelancer collaboration and compare performance with and without AI assistance.

Summary

This study investigates how generative AI tools can better support freelancer collaboration by conducting co-design sessions with 27 freelancers. Participants critiqued current AI tools for their lack of contextual understanding, risks of over-reliance, and threats to creative agency. They envisioned "auxiliary AI" systems that enhance collaboration without dominating it, emphasizing human-led workflows, productive friction, and support for diverse skill levels. Using DALL-E as a design probe, the study surfaced actionable design principles and highlighted the need for tools that preserve creative authenticity and foster equitable collaboration. These findings provide a foundation for reimagining generative AI as a facilitator of meaningful, human-centered teamwork.