This thesis explores how AI-powered tools can be designed to support designer-AI collaboration in user experience design, with a focus on professional practice in high-stakes contexts. While AI offers predictive and generative capabilities based on large-scale data, its use in safety-critical systems raises challenges related to opacity, over-reliance, and the erosion of creativity and skill development. In response, this thesis adopts a human-centered approach to explore designers’ needs, expectations, and engagement with AI in high-stakes contexts.

The thesis comprises five papers, including a systematic review and empirical studies in which expert designers co-designed and evaluated an AI-supported prototype. The synthesized findings were developed into a framework that highlights designer-AI collaboration as a situated and relational practice shaped by the interplay of context, designer characteristics, tool design, and AI-related elements. This thesis focuses on AI-supported data sensemaking, informing high-stakes design decisions with insights derived from user interaction data.

Designers’ needs, expertise, trust, and AI literacy influence how they interpret, assess, and interact with AI tools, while agency, information, and interaction enable effective, meaningful, and critical engagement with AI. This work contributes to HCI and design research by identifying the key elements and their interrelations that shape designer-AI collaboration in high-stakes contexts. It foregrounds the challenges of designing AI-supported tools that help designers remain grounded in real-world practice without diminishing contextual richness, while creating interaction paradigms that support flexible enactment and negotiation of agency. To operationalize these insights, this thesis presents a question bank to guide thoughtful inquiry and critical reflection in the design and development of human-centered and context-sensitive AI-supported design tools.