From Blank Box to Creative Partner: Designing Ecological On-Ramps for First-Time AI Artists

Paper Title

From Blank Box to Creative Partner: Designing Ecological On-Ramps for First-Time AI Artists

Publication Info

• Topic area: Human-Computer Interaction (HCI) and AI-enabled creativity support.
• Keywords: generative AI, creativity support tools, human-AI co-creation, ecological design, embodied interaction, longitudinal study, single-case ethnography, creative workflows, design patterns, attuned ambivalence.

Background and Problem

• Problem / challenge: Current HCI research lacks empirical understanding of how first-time users, particularly creative professionals, adopt generative AI tools. Existing studies focus on experienced users or isolated tool evaluations, leaving gaps in addressing challenges like blank-box paralysis, decision fatigue, and ownership anxieties.
• Significance: Understanding first-time adoption is critical as generative AI tools become increasingly accessible to millions of creative practitioners. This moment offers an opportunity to shape how these technologies integrate into creative practice.
• Motivation and related work: Prior research has explored AI co-creation systems, prompt engineering, and creativity support tools but has largely focused on experienced users. Studies on professional artists often rely on interviews, while tool evaluations emphasize efficiency and ideation throughput. This paper addresses the gap by focusing on the liminal moment of first-time AI adoption and its integration into existing creative ecologies.

Solution

• Proposed approach: A 10-week longitudinal ethnographic study of a professional artist’s first encounter with generative AI, emphasizing ecological and embodied practices.
• Novelty:
  1. Empirical insights into how a professional artist developed creative practice with AI, including concepts like situated prompt craft, elastic rhythms, and attuned ambivalence.
  2. Four actionable design patterns: park-and-resurface, warm-up modes, affect-to-action bridges, and comfort controls.
  3. Methodological contribution demonstrating the value of longitudinal, ecological studies for understanding human-AI interaction.
• Procedure and key techniques:
  • Observing one artist (“Anna”) over 10 weeks using a mix of diary studies, in-studio observations, interviews, and artefact collection.
  • Providing light scaffolds like LoRA fine-tuning, weekly check-ins, and a contact-sheet print pipeline.
  • Analyzing data through thematic coding, visual ethnography, and member checks to identify patterns of interaction and co-creation.

Results

• Concrete findings:
  • Interaction patterns emerged from material practices shaping digital tool use, not from mastering AI logic.
  • Temporal rhythms like park-and-resurface and elastic rhythms were essential for productive work.
  • Success was marked by productive friction, sustained ambivalence, and material interruptions rather than efficiency or output quality.
• Advantage over baselines:
  • The study challenges efficiency-first paradigms by demonstrating that deliberate friction and material engagement can foster creative success.
  • Observed practices like mood-first prompting and cross-surface workflows reveal alternative pathways to AI adoption.
• Experiments / evaluation:
  • The study involved one artist with 7+ years of professional practice, no prior AI experience, and access to custom model training.
  • Data included 32 hours of observed video, 22.5 hours of screen recordings, 400+ generated images, 68 Canva boards, and 100+ physical artefacts.
  • Indicators like resurfacing rates, cross-surface handoffs, and variance adjustments were used to analyze patterns.
• Limitations and future work:
  • Findings are based on a single case and may not generalize to other users or contexts.
  • The study reflects a privileged position (time, resources, technical support) and excludes novice creators or commercial artists under deadline pressure.
  • Future work should test the identified patterns with diverse user populations and explore comparative studies.

Summary

This paper presents a longitudinal case study of a professional artist’s first encounter with generative AI, revealing how creative success emerged through ecological resistance and material engagement rather than computational efficiency. Key contributions include three observed phenomena (situated prompt craft, elastic rhythms, attuned ambivalence), four actionable design patterns (park-and-resurface, warm-up modes, affect-to-action bridges, comfort controls), and speculative provocations challenging efficiency-first paradigms. While the findings are specific to one artist, they suggest alternative design spaces for AI-enabled creativity support tools that prioritize friction, ambivalence, and cross-surface workflows. Future research should explore these patterns in broader contexts to understand their applicability and impact.