De-skilling, Cognitive Offloading, and Misplaced Responsibilities: Potential Ironies of AI-Assisted Design
Citation. Shukla, P., Bui, P. N. N., Levy, S., Kowalski, M., Baigelenov, A., & Parsons, P. (2025). De-skilling, Cognitive Offloading, and Misplaced Responsibilities: Potential Ironies of AI-Assisted Design. Extended Abstracts of the CHI Conference on Human Factors in Computing Systems (CHI EA ‘25). https://doi.org/10.1145/3706599.3719931
Summary. This CHI EA paper examines how UX practitioners perceive GenAI in design work by analyzing public practitioner articles and UX-focused Reddit discussions. Practitioners saw AI as useful for repetitive work and ideation, but also raised concerns about over-reliance, cognitive offloading, de-skilling, hallucinations, bias, and the erosion of design judgment. The authors connect these concerns to classic human-automation literature, especially the ironies of automation and the substitution myth. Their core argument is that AI-assisted design should be treated as sociotechnical work redesign, not simple replacement of human tasks by machine strengths.
Key claims
- UX practitioners framed GenAI as a productivity tool for routine or repetitive UX tasks, including user-flow generation, research-script drafting, usability-testing workflow support, data analysis, pattern recognition, and other forms of design meta-work.
- Practitioners also described AI as a creative support or “second brain” for ideation, brainstorming, and early draft generation, while generally rejecting the idea that AI can replace human originality or intentionality.
- Human judgment remained central in practitioner discourse: designers emphasized empathy, contextual understanding, design rationale, hallucination checking, bias control, troubleshooting, and the ability to defend design choices.
- The paper argues that AI-assisted design risks repeating historical automation ironies: automation may reduce direct task execution while increasing monitoring, validation, troubleshooting, and responsibility for outputs the human did not produce.
- The authors warn against the substitution myth: dividing work into “humans are better at” and “machines are better at” fragments the system and ignores how automation changes human skills, responsibilities, procedures, and coordination.
- The paper identifies de-skilling and cognitive offloading as long-term risks if designers increasingly delegate brainstorming, problem framing, sketching, wireframing, and other foundational design activities to AI.
Evidence & limitations
- Evidence comes from qualitative content analysis of more than 120 public practitioner articles plus 62 UX-related Reddit posts and 1,575 comments from the prior three years.
- Four researchers used hybrid deductive and inductive thematic analysis, focusing this paper on productivity, creativity, and human judgment themes.
- The evidence is self-reported public discourse, not direct interviews, controlled observation, or longitudinal behavioral evidence of skill change.
- Some included online content may have been AI-generated or shaped by self-report and platform-selection biases.
- The paper captures practitioner perceptions at a fast-moving moment in GenAI adoption and cannot determine how attitudes or actual skill trajectories change over time.