Understanding Socio-technical Factors Configuring AI Non-Use in UX Work Practices

Research Background and Problem

• Issues or Challenges Identified by the Authors:

  • Artificial Intelligence (AI) tools are believed to revolutionize user experience (UX) workflows, but their actual adoption and use are highly complex. In many cases, AI may be unnecessary or even unwelcome.
  • Current research predominantly focuses on the use of AI while neglecting the phenomenon of "non-use." There is a lack of in-depth exploration into the reasons behind non-use.

• Why This Issue Is Important:

  • Studying non-use can reveal unmet user needs, potential concerns, and overlooked forms of work. Understanding non-use challenges the logic of technological solutionism, thereby providing a more comprehensive understanding of the boundaries of technology applicability.
  • In certain scenarios, traditional workflows may still be effective or even more appropriate, warranting attention alongside AI reliance.

• Research Motivation and Related Work:

  • Research in related fields often emphasizes the facilitation of work culture and practices by AI technologies, but there is insufficient focus on the phenomenon of AI non-use, its causes, and its multi-level connections.
  • Using the sociotechnical systems framework, the authors explore the dimensions and interactions of non-use in UX work, aiming to understand the complexity of technology adoption from a novel perspective.

Solution

• Proposed Methods or Solutions:

  • Conduct semi-structured interviews with 15 UX practitioners to study and summarize specific factors contributing to AI non-use, including constraints at the personal, professional, organizational, and societal levels.
  • Employ the sociotechnical assemblage framework to analyze non-use as the result of multi-layered, complex relationships rather than a singular choice.

• Innovative Aspects of the Solution:

  • Introduces "non-use" as a new perspective for studying AI technology adoption, moving beyond traditional research approaches that view technology use as the ultimate goal.
  • Highlights the dynamic role of multi-level constraints (e.g., professional values, organizational culture, legal regulations) in shaping non-use and incorporates these factors into considerations for technology design and policy participation.

• Implementation Steps and Key Techniques:

  1. Recruitment and Interviews: Conduct interviews with UX practitioners from diverse professional backgrounds, including design consulting, finance, healthcare, and streaming services.
  2. Data Analysis: Use inductive analysis to extract themes related to AI non-use from interview data and position the dynamic interactions of factors at different levels within the sociotechnical assemblage framework.
  3. Results Summary: Generate insights into the specific impacts of AI non-use on UX practices and propose recommendations for policy and tool design.

Research Outcomes

• Specific Findings:

  • Identified three main categories of non-use factors:
    1. Characteristics of UX Practice: Structured workflows, the driven nature of the design process, and the necessity of human judgment.
    2. Professional Values and Responsibilities: Concerns about AI-induced bias, inaccurate results, responsibilities for data security and privacy protection, and the need for designers to maintain autonomy.
    3. Organizational and Societal Levels: The influence of organizational decision-making, privacy laws, and client contracts on the adoption of AI tools.
  • Non-use is not a singular act of rejection but a dynamic, multi-layered, and sometimes contextualized practice.

• Advantages Over Existing Solutions:

  • Provides a comprehensive analysis of non-use, expanding the traditional focus of HCI research on AI tool usage.
  • Not only highlights the limitations of technological functionality but also emphasizes how policies and societal factors shape the conditions for non-use, which is less commonly addressed in technology-focused research.

• Experimental or Evaluation Results:

  • Interview participants generally indicated that AI tools have limitations in user research and human-computer interaction, such as understanding irrational user behavior, capturing subtle contextual cues, and performing multidimensional data analysis.
  • Data privacy, regulatory policies, and internal organizational decisions significantly influenced the formation of non-use.

• Limitations and Future Directions:

  • Limitations: The study primarily reflects the perspectives of practitioners in North America and South Korea, potentially overlooking diversity in other cultural or contextual backgrounds. It also does not deeply explore how participants' social identities (e.g., race, class, gender) influence non-use practices.
  • Future Directions:
    • Investigate how diverse cultural contexts shape AI non-use and extend research to freelancers and other non-traditional work structures.
    • Explore design practices that support the deliberate rejection of AI tools and develop toolkits to encourage reflection on the boundaries of technology applicability.
    • Conduct in-depth studies on how policies and technology design can collaborate to effectively regulate the boundaries of technology use and non-use.

Through the above analysis, it is evident that this paper not only enriches the understanding of the dynamics of AI technology usage but also provides important guidance for AI tool design, evaluation, and industry practices. This has profound implications for policymakers, technology designers, and UX professionals.