Understanding User Perceptions and the Role of AI Image Generators in Image Creation Workflows

Research Background and Issues

• What problems or challenges did the authors identify?
  Existing generative AI tools have been extensively explored in both theory and practice, but there remains a gap in understanding how AI image generators (AIGs) specifically support users in the image creation workflow, particularly from the initial creative concept to post-production sharing. Key issues include:

  • A mismatch between the functionalities supported by current AIGs and users' intentions and expected outcomes.
  • Insufficient research on social and emotional factors, such as motivations for sharing generated images and their potential impacts.

• Why is this issue important?
  AIGs not only transform the technical methods of creative design but also profoundly influence user behavior and social interactions. A deeper understanding and optimization of these tools' usage could lead to more efficient and convenient creative support, as well as enhanced social connection value.

• Research Motivation and Related Work
  While some studies have explored the application of generative AI in the field of human-computer interaction (HCI), most analyses focus on specific platforms (e.g., Midjourney or DALL-E) and lack systematic analysis of multiple generators in real-world scenarios. Additionally, most research overlooks the specific roles of generated images at different stages of the creative workflow, as well as users' emotional and sharing behaviors.

Solutions

• What methods or solutions did the authors propose?
  Through in-depth interviews with 26 AIG users and "think-aloud" tasks, the authors collected and analyzed user behavior patterns, decision-making processes, and social interactions across multiple generators. They proposed a classification model with four usage scenarios (strategic, discovery, interest, and explorational-oriented) and explored the roles of generators throughout the workflow, from concept development to final image creation.

• What is innovative about this solution?

  1. Full workflow analysis: Unlike previous studies that adopt a single-tool or single-task perspective, this study uses a six-stage creative workflow model to understand the specific roles of AIGs.
  2. User types and emotional insights: Beyond tool functionalities, the study reveals user behavior patterns and motivations for social sharing.
  3. Cross-tool comparison and integration: The research covers 14 different AIGs, offering diverse insights into tool comparison and integration strategies.

• What are the implementation steps and key technologies used?

  • Research Methods: The authors employed semi-structured interviews and "think-aloud" image creation tasks to collect data, followed by qualitative coding analysis. They categorized the data based on key models, such as Botella's six-stage creative model.
  • Data Analysis Framework: The study collected and integrated information on tool selection, input strategies (e.g., simplifying or refining text prompts), evaluation of generated results, and subsequent actions (e.g., adjustments and sharing).

Research Findings

• What specific findings were achieved?

  1. Four usage scenarios: The authors categorized the usage cases of 26 participants into strategic-oriented, discovery-oriented, interest-oriented, and explorational-oriented scenarios. These scenarios differ significantly in terms of goals, clarity of initial vision, and tool application stages.
  2. Decision-making process: The study summarized three key decisions users make during image creation: vision development and prompt input, evaluation of generated results, and adjustment/refinement processes.
  3. Social impact: The research found that users' sharing behaviors after generating images created additional social value, such as fostering communication or building social relationships through images.

• What advantages does it have compared to existing solutions?

  • Compared to studies focused on single tools, this research is more comprehensive due to its multi-platform and multi-generator comparison.
  • It integrates technical functionalities with social interactions, expanding the understanding of AIGs' potential as social tools.

• What were the experimental or evaluation results?

  • In explorational-oriented scenarios, users tend to focus on the capabilities of generators, while in strategic-oriented scenarios, they emphasize controlling the details of generated results.
  • Generated images have shown potential as tools for social communication (e.g., conversation starters or enhancing collaboration) in certain scenarios, such as interest-oriented use cases.

• Limitations and Future Directions

  • Limitations:
    1. The data sample is skewed toward academic environments (student users), which may limit the diversity of perspectives from professional users in industry settings.
    2. The study lacks observations of AIG usage in dynamic and long-term contexts.
  • Future Directions:
    • Explore usage scenarios in more professional fields (e.g., artists or designers).
    • Monitor long-term changes in user behavior as generative AI technology advances.
    • Enhance user support for prompt input, such as optimizing historical records and multimodal input capabilities.

Conclusion

This study innovatively explores the impact of AI image generators on the image creation workflow from both functional and social perspectives. The classification model and decision-making process summaries provide valuable references for practical design and HCI research. Additionally, the authors propose design recommendations, such as iterative comparison and tool integration, which can help generative AI technologies more effectively support the creative needs of diverse user groups.