Abstract:

Doctoral education in Norway is characterised by a paradox: PhD candidates are generally well funded through scholarships and institutional support, yet completion rates and time to degree remain a persistent challenge. The average age at doctoral defence is approximately 38 years, indicating extended doctoral trajectories. Empirical data show that only around 64% completed within six years (Norwegian Directorate for Higher Education and Skills). These figures persist despite comparatively favourable funding conditions, suggesting that work organisation and structural “time thieves” play a critical role.

Research indicates that PhD candidates spend a substantial proportion of their working time on administrative and coordination-related tasks. Survey-based studies suggest that up to one-third of doctoral working hours may be devoted to documentation, reporting, email correspondence, scheduling, and compliance with digital systems. While necessary for institutional functioning, such tasks are epistemically low in meaning-making value. When administrative labour consumes disproportionate time, the PhD role risks shifting from knowledge production to knowledge administration, with implications for research quality, motivation, and academic identity.

This paper explores whether responsible uses of voice-activated artificial intelligence (AI) assistants may help mitigate such time thieves and enhance completion capacity and health empowerment. Doctoral work is cognitively demanding, and prolonged screen-based activity combined with fine-motor strain and performance pressure poses physical and psychological challenges. Musculoskeletal discomfort in the neck, shoulders, wrists, and hands is widely reported, alongside stress-related symptoms linked to fragmented workdays.

The study adopts a mixed methods design combining quantitative survey data (N = 62), an intervention study, and qualitative interviews. The empirical material is drawn from the Nordic Research School in Educational Research (NORED), characterised by international cohorts and extensive digital infrastructures. The survey mapped time use, administrative workload, physical strain, and attitudes towards AI. Results indicate widespread experiences of time scarcity, frequent musculoskeletal discomfort, and cautious but generally positive attitudes towards AI, particularly for non-epistemic core tasks.

Based on these findings, an intervention was designed in which PhD candidates integrated voice-activated, multilingual AI assistants into selected administrative and preparatory academic tasks over a defined period. These included drafting routine texts, summarising documents, managing emails, organising notes, and interacting with administrative systems through speech-to-text interfaces. Administrative uses were prioritised as low-risk areas for AI implementation that do not threaten research integrity or disciplinary standards.

Analytically, the findings are discussed through a holistic framework integrating educational research, ergonomics, AI studies, and research leadership. The paper argues that voice-activated AI assistants, when used transparently and ethically, represent a low-risk, health-promoting innovation in doctoral education. Rather than undermining academic ideals, such use may help restore balance between necessary administrative labour and core knowledge production.

Keywords:

Doctoral education, completion rates, AI, multimodality, Health empowerment.