Impact on Vulnerable Groups and Individual Technical Factors. The use of AI to predict falls in hospitalized patients may exacerbate inequalities due to biases present in training data. These biases affect model fairness, potentially leading to inadequate care for marginalized groups or vulnerable populations and increasing the risk of errors in prediction and care management (Chen & Xu, 2023).

Technological and Economic Barriers. Unequal access to advanced technologies such as AI and virtual reality in nursing education puts under-resourced institutions at a disadvantage. This limits their students’ ability to develop critical skills, creating disparities in the quality of education received (Harmon et al., 2021).

Overreliance on AI Outputs. While AI has improved diagnostic accuracy and streamlined clinical processes, excessive reliance on these technological models may erode professionals’ ability to make decisions grounded in clinical judgment. This becomes problematic in complex cases where AI may fail to provide adequate solutions due to limitations in data or model training (Huqh et al., 2022).

Loss of Critical Skills. In nursing, the use of AI may limit professionals’ ability to address complex cases that do not fit predefined models (Ibuki et al., 2024). This technological dependence increases the risk of losing essential skills such as empathy and adaptability, which are crucial for providing personalized care (Ibuki et al., 2024).

Bias in Algorithm Development. Although there are notable examples of AI systems developed in collaboration with healthcare professionals, such as IBM Watson for Oncology (Somashekhar et al., 2018), which aim to address specific clinical needs, several studies highlight that such collaboration is not always the norm (Abbasgholizadeh Rahimi et al., 2021; Abràmoff et al., 2023). In some instances, algorithms are trained without sufficient understanding of the clinical context, leading to outputs that perpetuate existing biases. This underscores the importance of ensuring consistent and meaningful integration of healthcare personnel throughout the design and implementation phases of AI technologies.

Variability in Accuracy. Limitations in data generalization across clinical contexts have led to false positives and negatives, which may compromise clinical decision-making. Additionally, variability in accuracy is influenced by factors such as input data quality and imaging conditions (Li et al., 2021; Liu et al., 2019). This concern is especially critical in pediatric intensive care settings, where accuracy in predicting complications, such as venous thrombosis, is essential for patient safety (Lei et al., 2023; Li et al., 2021).

Disparity in Data Collection. The quality and diversity of datasets used to train AI models is another key contributor to algorithmic bias. Systems trained on data from specific populations often underrepresent minorities, thereby perpetuating disparities in care. This issue becomes especially problematic when attempting to predict social determinants of health, as historical data fail to adequately reflect marginalized communities (Ronquillo et al., 2022).

Reduction in Human Supervision and Skills. The adoption of technologies such as digital angiography in hemodialysis patients and AI in medical education has raised concerns about diminishing human involvement in healthcare. Excessive reliance on technology may reduce clinical supervision and hinder the development of essential human skills such as empathy and communication. As a result, future healthcare professionals may become less capable of addressing the emotional and ethical dimensions of care, due to an overemphasis on technological solutions (Lee, Wu et al., 2021; Mi, 2022).

Weakening of Patient–Provider Relationships. Patients express concern that technology could replace human interaction, undermining trust and the quality of their relationships with healthcare professionals. Moreover, inadequate training in AI use increases the risk of errors and limits the effectiveness of these tools, potentially compromising the quality of care (Fazakarley et al., 2024).

Data Processing and Protection. AI systems process vast amounts of personal data, raising concerns about how this information is protected and utilized. In many cases, patients are not fully informed about how their data are collected and processed, which may compromise their privacy (Ng et al., 2022). This is particularly critical in direct-to-consumer health applications, where lack of transparency regarding the use of personal data can result in serious vulnerabilities (He et al., 2023).

Challenges in Interoperability and Security. The use of AI chatbots in programs such as weight loss interventions has demonstrated that integration with multiple devices and platforms can heighten the risks to personal data security. Users may be unaware of the implications of such integration, leaving them vulnerable to privacy breaches if the platforms involved do not implement adequate safeguards (Chew, 2022).

Transparency in Decision-Making and Accountability. The opacity of AI systems prevents healthcare professionals and patients from understanding how clinical decisions are made, complicating accountability in the event of harm or error (Masoumian Hosseini et al., 2023). In critical situations, such as emergency care, the absence of clear explanations from AI systems may erode trust and limit human intervention, worsening the impact of potential failures (Barwise et al., 2024; Masoumian Hosseini et al., 2023).

Need for Clear Accountability Mechanisms. There is a pressing need to clearly define responsibility in cases of error. Automated systems often make decisions without direct professional input, making it difficult to assign accountability when problems arise (Barwise et al., 2024). This highlights the importance of establishing regulatory frameworks and oversight mechanisms to determine who should be held accountable—whether technology developers, healthcare providers, or both.

Reduction in Human Interaction. The automation of processes such as pain assessment through AI may reduce direct interaction between healthcare professionals and patients, potentially dehumanizing care by limiting empathy and negatively affecting the quality of the service provided. Furthermore, when AI models lack representative data, their effectiveness across different clinical settings may be compromised, raising concerns about their universal applicability (Abuzaid et al., 2022; Zhang et al., 2023).

Job Uncertainty. Automation raises concerns about its impact on employment. As more tasks become automated, the demand for personnel may decline, creating uncertainty regarding job stability. This issue has been observed in other sectors, where workers experience high stress levels due to job insecurity caused by automation. Moreover, the psychosocial factors that affect worker well-being are not always considered (Abuzaid et al., 2022; Cheng et al., 2021).

Table 2 shows how studies were grouped by thematic categories related to the phenomenon of interest

Table 2
Risks of AI in Nursing Care

Source: Own work.

Optimization of Clinical Resources and Workflows. In certain settings, AI has proven effective in optimizing workflows and resource management in clinical care. From automating documentation and creating personalized care plans to predicting the risk of falls and pressure ulcers, AI has enabled nurses to focus more on direct patient care, thereby improving the quality of service (Ng et al., 2022). In hospital management, AI has facilitated the prediction of length of stay and mortality in patients with diabetes and hypertension, optimizing bed utilization and other clinical resources (Barsasella et al., 2022).

Support for Clinical Decision-Making. AI has been shown to support clinical decision-making in areas such as hospital readmission prediction and respiratory infection management. Its implementation has significantly reduced rehospitalization rates, improving clinical outcomes (Romero-Brufau et al., 2020). In addition, some systems have assisted in patient triage by offering monitoring intervals based on clinical guidelines, thereby improving diagnostic accuracy (Li et al., 2022). In cases involving language barriers, AI has enhanced equity in care by quickly identifying patients who require interpretation services (Barwise et al., 2024).

Reduction of Workload. In the nursing field, AI has helped reduce workload by automating routine tasks such as intravenous bag monitoring, pressure injury management, and administrative duties, freeing up time for other responsibilities (Hwang et al., 2023; Chen et al., 2022). The introduction of robots for repetitive tasks has reduced physical exhaustion, allowing professionals to focus on clinical judgment and empathetic care (Zrínyi et al., 2022).

Prediction of Complications. AI has been essential in predicting complications. In diabetic patients, AI models identified risks of neuropathy, nephropathy, retinopathy, and amputations, enabling personalized interventions (Gosak et al., 2022; Mousa et al., 2023). In hospitalized patients, AI predicted severe hypoglycemia and the risk of developing pressure injuries, facilitating early identification of adverse events (Fralick et al., 2021; Pei et al., 2023).

Real-Time Monitoring. AI has also enhanced real-time monitoring in various areas. In emergency departments, algorithms predicted complications in patients experiencing hyperglycemic crises (Hsu et al., 2023).

Support for Mental Health and Treatment Adherence. In the mental health field, AI accurately interpreted human emotions, aiding in diagnosis and emotional support (Elyoseph et al., 2024). AI models predicted depressive symptoms in older adults without requiring questionnaires (Susanty et al., 2023). Moreover, AI helped monitor treatment adherence through smartwatches, ensuring proper medication intake (Odhiambo et al., 2023).

The integration of AI and virtual reality has been shown to enhance pain management education, allowing students to practice in simulated environments without posing risks to patients. Furthermore, AI personalizes learning scenarios, improving knowledge retention (Harmon et al., 2021). In clinical practice, AI has supported evidence-based decision-making, promoting continuous education for nursing professionals and helping them stay current with recent practices and technologies (Abuzaid et al., 2022). In addition, the use of humanoid robots in hospital settings has also proven effective in promoting health literacy and vaccine comprehension (McIntosh et al., 2022).

Continuous and Emotional Care. AI and autonomous robots have transformed the care of patients with chronic illnesses and advanced dementia, demonstrating reductions in agitation, improved emotional well-being, and more patient-centered attention to emotional and social needs (David et al., 2021). These technologies also automate physical tasks such as mobilization, allowing staff to devote more time to specialized care (Cai et al., 2021; Stokes & Palmer, 2020).

Personalized Risk Prediction. AI has shown efficacy in clinical risk prevention. In nursing homes, predictive models identified key factors for preventing pressure ulcers (Lee, Shin et al., 2021). In home care, AI has enhanced hospitalization prediction by analyzing clinical notes (Topaz et al., 2020). AI also optimizes remote monitoring of older adults, detecting falls and cognitive decline, and supports health coaching for patients with type 2 diabetes as well as interventions for autism spectrum disorder (Schütz et al., 2022; Di et al., 2022; Jia et al., 2023).

AI has significantly improved access to critical information in hospital settings, enabling more informed and efficient decision-making in high-pressure situations. For example, AI-powered chatbots in hospitals have provided quick and accurate responses to caregiver inquiries, reducing errors and workload (Daniel et al., 2022). In emergency departments, AI-assisted systems for real-time medical record entry (via voice-to-text data conversion) have improved triage efficiency and accuracy by capturing more relevant information for patient care (Cho et al., 2022).

The integration of AI into healthcare has created new opportunities for interdisciplinary collaboration, as nurses, software engineers, and other professionals work together to develop more optimized technologies (Shang, 2021). Involving patients and the public in the design of these tools ensures their functionality while fostering more inclusive and equitable care that adapts to diverse user needs (Zidaru et al., 2021).

The findings mentioned above are summarized in Table 3. The studies were grouped into thematic categories related to the phenomenon of interest.

Table 3
Opportunities for AI in Nursing Care

Source: Own work.