Healthcare systems worldwide are betting heavily on artificial intelligence to solve a nursing workforce crisis that shows no signs of easing — but whether these tools actually lighten the burden on nurses, or simply transform one form of strain into another, has remained largely unresolved. This systematic review, registered in PROSPERO and conducted across six databases, offers the most structured synthesis to date on that question.
Drawing on 20 studies from eight countries published between 2010 and 2025, the review examined how AI-enabled workflows — including telehealth triage systems, clinical decision support platforms, and automated documentation tools — affect nursing workload, emotional well-being, job satisfaction, and workforce sustainability. The workload picture was decidedly mixed: roughly half the studies (10 of 20) reported reductions in workload, while five recorded mixed effects, two showed no meaningful difference, and two found workload actually increased. Emotional well-being and job satisfaction outcomes were similarly heterogeneous, with organizational culture and implementation quality emerging as key moderating variables rather than the technology itself.
This finding deserves careful interpretation. The concept of "technostress" — cognitive overload arising from the continuous adaptation demands of digital tools — has been documented in nursing contexts for over a decade, and this review confirms it remains a live concern even with sophisticated AI systems. The paradox is structural: the same automation intended to reduce charting burden or triage time can simultaneously fragment clinical judgment, erode professional autonomy, or introduce new alert-fatigue dynamics. The heterogeneity across studies also reflects a fundamental methodological problem — AI "tools" span an enormous functional range, making aggregated conclusions fragile. The review's 20-study scope, while meaningful for a PRISMA-compliant synthesis, limits statistical power and subgroup analyses. This is confirmatory rather than paradigm-shifting evidence, but it does underscore that deployment context — staffing ratios, training adequacy, institutional culture — likely determines outcomes more than the technology itself.