Generative AI is rapidly becoming a key part of knowledge work, offering opportunities to increase productivity, creativity, job satisfaction, and improve learning outcomes, while also raising concerns about potential skill erosion as tasks are increasingly delegated to GenAI systems. Despite growing research on productivity, ethics, and governance, limited attention has been given to how knowledge workers manage their interaction with GenAI to sustain professional skills, which are critical for personal development, organizational efficacy, and societal resilience.

This thesis aims to enhance understanding of how GenAI impacts professional skills and whether knowledge workers self-regulate their use to counteract skill changes. Using a qualitative research approach, 38 semi-structured interviews were conducted with knowledge workers across ten sectors. Thematic analysis revealed that GenAI use can result in four distinct skill outcomes: development, maintenance, erosion, and revaluation. Crucially, these outcomes depend on the way workers engage with GenAI rather than the technology alone, with skill revaluation showing that erosion is not always perceived as harmful.

Knowledge workers employ self-regulatory strategies before, during, and after GenAI use, including setting boundaries, tailoring usage to specific situations, critically reviewing outputs, and rewriting or documenting results. Engagement in self-regulation is shaped by drivers across interactional, organizational, and societal contexts, yet organizational support for mitigating risks of skill erosion remains limited, leaving responsibility with individual workers.

By integrating insights on skill outcomes, strategies, and drivers, this study highlights how self-regulation unfolds within a complex sociotechnical system. The findings provide implications for knowledge workers, organizations, and policymakers to design interventions that foster responsible adoption of GenAI, protect professional competencies, and strengthen societal resilience. Future research should further explore organizational-level interventions, governance structures, and profession-specific experiences to ensure skill preservation in the age of GenAI.

This thesis explores how data-enabled design can be used to create personalized care pathways for chronic patients while reducing the burden on healthcare providers. The research focuses on three diseases colorectal cancer, pulmonary fibrosis, and sarcoidosis, and investigates how unstructured patient narratives from online peer-support communities can be transformed into clinically relevant behavioral insights.

Over 13,000 anonymized patient posts were collected, pre-processed, and analyzed using a GPT-4o-assisted semantic clustering and verification pipeline. A multi-step methodology combined natural language processing, chain-of-verification prompting, behavioral science frameworks, and manual review to generate detailed patient behavior profiles. These profiles captured goals, motivations, and challenges, and were validated through co-creation sessions with clinicians from Erasmus MC. The sessions refined the outputs and mapped patient behavior across different stages of the care journey. These behaviors were further thematically clustered into eight cross-condition patient archetypes.

These archetypes informed the design of two complementary interventions. The first is a patient-facing mobile application that supports goal setting based on quality of life, mood and symptom journaling, and symbolic peer support. The second is a clinician-facing dashboard, integrated into the HIX electronic health record, that visualizes behavioral trends alongside clinical data and generates AI-assisted care modules for review and assignment.

The outcomes include a reproducible data-to-design pipeline, validated behavioral archetypes, high-fidelity prototypes of patient and clinician tools, and a Vision 2040 roadmap for phased national implementation. This work demonstrates that patient-authored narratives contain actionable behavioral signals that, when combined with clinical expertise, can guide adaptive, behavior-aware care planning. The thesis contributes a practical framework for embedding these insights into decision-support systems that strengthen personalization, preserve clinician autonomy, and enhance patient engagement at scale.

This study investigates whether Large Language Models (LLMs) exhibit victim blaming tendencies when analyzing traffic accident scenarios. Using a systematic factorial design, 144 road safety scenarios were tested across ChatGPT-4o and DeepSeek-V3, varying risk behavior, demographics, injury severity, and driving context. Through mixed-methods analysis of 288 responses, the research reveals that LLMs do not exhibit traditional victim blaming but demonstrate compliance-based attribution. They adapt their analytical approach based on question framing rather than maintaining consistent safety principles. When asked about prevention, models provided comprehensive systems analysis with 89.5% of suggestions targeting systemic factors. However, responsibility attribution followed context-driven patterns: 100% driver blame in private scenarios versus 69.4% company blame in work-related scenarios, regardless of demographics. The convergence of these patterns across different AI architectures suggests fundamental challenges in analytical consistency rather than model-specific biases. These findings have critical implications for AI deployment in safety-critical contexts, revealing that sophisticated responses can mask inappropriate analytical frameworks. The study contributes new theoretical concepts to AI ethics and provides evidence-based guidance for ensuring AI systems support rather than undermine advances in safety science.

Chronic cardiovascular diseases (CVD) represent a growing global health burden, where therapeutic compliance is critical for effective disease management and prevention of complications. Mobile health (mHealth) applications have emerged as promising tools to support patients in managing chronic conditions, yet their adoption remains inconsistent. Existing technology adoption models often fail to fully capture the psychological dimensions that influence patient decisions. This study addresses this gap by investigating how health beliefs shape the adoption of mHealth applications for therapeutic compliance among chronic CVD patients. The research integrates the Health Belief Model (HBM) and the Technology Readiness and Acceptance Model (TRAM) to explore both psychological health beliefs and technology readiness traits in tandem. The study adopts a qualitative, exploratory design, employing semi-structured interviews with 11 CVD patients in the long-term management phase. Thematic analysis, supported by ATLAS.ti, was used to develop three key themes from participant narratives.