Union negotiations traditionally focus on structural modifications that address clinician demands. However, amid the backdrop of workforce shortages, AI technologies like predictive staffing and virtual nursing have been proposed to streamline repetitive tasks and limit errors to reduce nurses’ daily workloads.

Automated scheduling systems, for example, are designed to predict nurse schedules based on historical admission data and changing patient volumes. Machine learning algorithms can identify patterns that may otherwise be overlooked by human nurses, such as the effects of weather or local events on emergency department traffic.

To reduce the burden often associated with nursing documentation, voice-to-text and algorithm-driven triage technologies have been widely adopted. Enabled by natural language processing (NLP), a type of AI technology, ambient scribes and automated charting assistants listen to and record interactions between nurses and patients in real time. Ambient voice technologies eliminate the need for manually entering patient data, which may help ensure that clinically relevant details are retained in patient records.

For nurses experiencing staffing pressures and greater workloads, these technologies have the potential to reduce the burden of clinical practice and free time for direct patient care. Joe-Ann Fergus, RN, PhD, Executive Director at the Massachusetts Nurses Association, acknowledges the urgency behind the search for innovative solutions. “People who are practicing right now are desperate for anything that will give relief. There is always a level of, ‘maybe this will be the thing.’”

Workforce shortages highlighted by recent nursing strikes have also increased pressure on nursing programs to explore new approaches for preparing future clinicians. Virtual simulation platforms, which became more widely adopted during the COVID-19 pandemic, provide immersive learning experiences to repeatedly practice procedures in a controlled setting.

Simulation-based training allows students to make mistakes in a low-stress learning environment without risking patient safety. The use of AI-based virtual patients in these simulations teaches nursing students how to make clinical decisions in realistic scenarios. Generative AI models like ChatGPT, Copilot, and Gemini are also being used to enhance engagement in the classroom by personalizing explanations, analogies, and examples that address individual learning needs and preferences while providing real-time feedback on student performance.

In today’s postpandemic environment, a new generation of clinicians has been extensively trained through virtual simulation rather than through traditional, hands-on clinical experience. Although these technologies reduce clinician anxiety during training, questions remain about how effectively simulation alone prepares nurses for the unpredictable, high-pressure realities of the hospital.

“This is the nexus point between AI and technology in health care,” warns Fergus. “When clinicians are deskilled and cannot truly question the output, it creates a risk of liability for both the patient and the clinician.”

This trend poses the risk of automation bias, which occurs when humans increasingly rely on automated systems to make crucial decisions. Automation bias reduces situational awareness, a crucial skill that clinicians develop to recognize how a patient’s condition is changing and anticipate what may happen next. Over time, this erosion of baseline skill development may compromise patient care, as well as the long-term effectiveness of these AI-driven tools.

To ensure the efficiency of these systems without sacrificing their reliability, it is crucial for nurses to be involved in the design and development of clinical AI models. Nurses bring firsthand knowledge of decision-making workflows and the challenges that arise during the real-world implementation of new systems. Without this input, digital tools may fail to address the problems they are supposed to solve or introduce new burdens into already strained clinical environments.

Many health care systems are also adopting human-in-the-loop approaches that require nurses to verify AI-generated recommendations before they are implemented. Other organizations are requiring nurses to manually review and cross-check AI-generated alerts before administering patient care. Although necessary, this creates an additional layer of obligation for nurses to monitor for AI hallucinations and other subtle errors rather than offloading responsibilities as originally intended.

Ultimately, AI alone cannot resolve the workforce challenges highlighted by recent nursing strikes. The successful implementation of AI-enabled technologies in the clinic will depend on their technical performance, as well as their ability to meaningfully improve working conditions for nurses. “The solution isn’t just adopting AI,” Fergus notes, “it’s making sure it solves the problems nurses actually face.”