AI in Nursing

Daniella Topol, MS ’24, knew she faced a steep learning curve when she decided to enter Columbia Nursing’s Masters Direct Entry (MDE) program in 2023. Not only had she spent the previous 25 years as a creative director in the theater world, but she hadn’t taken a science course since high school. Moreover, she had no knowledge of generative artificial intelligence (AI), a tool that clinicians increasingly use. “I had to ask my 10-year-old daughter to show me how to use ChatGPT,” Topol says with a chuckle.

But today, as a student in the school’s Family Nurse Practitioner Doctor of Nursing Practice (DNP) Program, Topol uses ChatGPT regularly to test herself on material in her courses. She also sits on Columbia Nursing’s Student Advisory Council (SAC) on AI, helping to educate her classmates and others in the school community about the potential advances—and risks—that this emerging technology brings to the field. “I’m learning about AI for the first time, so playing a role in some small way with other students to explore what’s possible when integrating it into nursing education is really thrilling,” she says.

The establishment of the SAC came soon after the creation of the school’s Office of AI (OAI) in July 2024. Stephen Ferrara, DNP, a professor of nursing and the school’s associate dean of AI, directs the OAI and leads its efforts to familiarize students, faculty, and staff with the technology. “I’m working with students and faculty to help them understand AI as a tool that can be used in the classroom or the simulation lab and how we can deploy it in health care in a meaningful way,” says Ferrara.

Members of the SAC range from students with no prior experience in nursing to doctoral candidates. Together, they work to raise community awareness of AI; discuss its impact on students’ educational, research, and clinical experiences; and invite their peers to share their concerns about the technology. “We’re helping to represent students and move the conversation forward,” Topol says. The SAC has already held an informational town hall, an AI arcade, and a poster presentation to introduce students and nurses to the technology and also conducted a follow-up survey. “There is a lot of curiosity about integrating AI into nursing education and practice,” Topol says, including “how we harness it and what are its blind spots.” “For members of the SAC to be a resource for other students and members of the nursing community is key,” she adds.

What is AI? 

IBM, the multinational technology company, defines AI as “technology that enables computers and machines to simulate human learning, comprehension, problem solving, decision making, creativity, and autonomy.” It does this by collecting and processing massive amounts of data to create software algorithms that analyze and generate information. “Generative AI, like ChatGPT, has consumed every article that’s out there and uses it as an algorithm to predict what text you want it to give you in response to the questions you ask,” Ferrara says. “You put in a prompt, and it gives you an answer, or it predicts what should be written next in response to the question or information you type in.”

As a health care tool, AI is trained to extract and analyze large quantities of data from sources like patients’ electronic health records (EHRs), including descriptions of symptoms, lab findings, imaging results, or vital signs. It processes this data to detect unusual and potentially significant patterns or risks that clinicians might miss, like a patient’s risk of falling or deteriorating. Yet its capacity to predict patient outcomes is still experimental. “It’s very much emerging and not ready for prime time,” Ferrara notes.

One form of generative AI with practical bedside use that has emerged is the so-called “virtual scribe,” which records and transcribes patient-nurse conversations and then extracts important information from them. This spares nurses from having to continually return to their computers to write down the details of every patient interaction. “A lot of whatwe do in health care is documentation—filling out forms or other things that take up time but don’t have direct impact on patients,” Ferrara says. “AI will allow us to be more present and spend more time with our patients because we don’t have to spend all the time doing busywork. If we can find ways to better care for our patients, then that’s something we want to be researching and using.”

A tool for teaching and learning

These days, Columbia Nursing’s focus is on using AI as a teaching and learning tool, which it is beginning to integrate into its curriculum, says Mary Hickey, EdD, the vice dean of education and a professor of nursing. “Clinical applications of AI are continuing to evolve,” Hickey says. “It is being used to review images and scans, and aid in diagnosis. Although its full potential to improve patient outcomes has not been realized, we plan to integrate its clinical application into the curriculum as we learn more.” 

Meanwhile, students and faculty are gradually integrating AI, largely ChatGPT and other generative or large language models, into nursing education. For instance, some faculty members use it to help students learn content, interpret symptoms, create care plans, or improve writing skills. Some faculty are using AI to generate case scenarios. “Our Simulation Center has piloted virtual reality for skill building in some advanced practice RN programs,” Hickey says. “Students need to know how AI can help them make differential diagnoses, analyze reports, or interpret computer-generated analyses of scans.”

Students’ familiarity with the technology varies. For example, MDE student Pawel Maslag arrived at Columbia Nursing fluent in the basics of ChatGPT. He now uses it to better understand the material in his courses and to study by creating practice quizzes with questions. A longtime social activist and tutor, Maslag volunteered for nonprofits like AmeriCorps, Habitat for Humanity, and the U.S. Fund for UNICEF before enrolling in nursing school. He also launched an AI social impact lab to teach neighbors in his New Jersey community how to use the technology in their daily lives. Now a member of the SAC, Maslag—who earned his undergraduate degree in sociology from Columbia University in 2018—is eager to use his AI expertise and tutoring skills to help his current classmates. “I hope to bring my enthusiasm for AI literacy to the SAC, so I can advocate to make AI part of our curriculum,” he says.

Topol, who is working per diem as a registered nurse (RN) while earning her doctorate, also uses AI as a study aid. “I’ll ask it to give me a scenario where a patient has an abdominal issue so I can explore various diagnoses and treatments,” she says. She also hopes that AI will help her navigate the numerous databases, like PubMed, that can be used to research clinical questions. “I’m always wondering, ‘Do I go to this or that app?’” she says. “As a new RN and emerging provider, I don’t want to miss anything.”

PhD student Jiaojiao Wright, MS, calls generative AI a “mind-blowing” educational tool that has helped her grasp statistics in her doctoral studies. Born and raised in Hebi City, China, Wright earned a degree in English literature and linguistics in 2013 to pursue her goal of becoming a medical interpreter. After emigrating to the United States with her husband, she earned her master’s degree in nursing and began working as an RN for a hospital in lower Manhattan during the COVID pandemic. “The area has a huge Chinese population, and I saw so much inequity for people who don’t speak English, especially in the emergency room,” she says. To address this inequity, Wright devised a communication card system, featuring pictures that illustrated common phrases like, “I need… [the bathroom, water, a blanket],” to help Chinesespeaking patients and English-speaking providers communicate with each other.

The project earned her a grant for further study and led to a poster she presented in April at the 2025 annual scientific sessions of the Eastern Nursing Research Society. It also inspired her to pursue a career in nursing research.

As part of her doctoral work, Wright plans to use AI to enhance her communication card system. She already uses the technology to help her understand statistics. “Our statistics class is remote, so we can’t always get answers to our questions instantly,” she says. “But if I ask ChatGPT ‘What is a P value?’ it gives the answer right away and makes it simple. It can also explain logically and scientifically what a type 1 or type 2 error is. It is an important tool in my education.”

Because AI is evolving so quickly, it is imperative for all its users to understand its changes and implement strategic initiatives to ensure its meaningful adoption, Ferrara says. “We see AI being interwoven throughout the school to strengthen what Columbia does best: educate future nurses, deliver excellence in practice, and support worldclass research.”

AI in Research

In fact, Columbia’s entry into the world of AI is not brand new. The school has been training doctoral and postdoctoral research students in computational applications, under the rubric of informatics, for nearly a quarter of a century. It is thus the leading nursing school in the area of AI research in the world, according to Suzanne Bakken, PhD, vice dean for research, the alumni professor of nursing, and a professor of biomedical informatics. “For us, AI is not a method to help us do research, it is the focus of the research,” Bakken says. “We have been training PhDs and postdocs in AI for 23 years. It is why we ranked first in the bibliometric analysis of 2000-24.”

Columbia’s AI-related research features explorations of the predictive ability of the technology to improve patient outcomes. For example, one study is looking at the effectiveness of an AI tool that uses recorded communication between nurses and patients to identify high-risk patients during hospital-to-homecare transitions. Another is analyzing whether AI-recorded voices and speech are better than text at predicting whether patients have subtle cognitive deficits signaling the early stages of Alzheimer’s disease and related types of dementia. 

Sergine Delma, a student in the Family Nurse Practitioner DNP Program, is exploring yet another predictive application of AI: its potential to anticipate in primary care settings the progression of type 2 diabetes mellitus and comorbidities associated with the disease; use patient-specific data to personalize treatment plans; and use voice-based conversational AI to support self-management. “AI has been shown to help manage type 2 diabetes by reminding patients to check their insulin and blood sugars,” says Delma. “These tools can act as powerful equalizers when used thoughtfully, by not only enhancing clinical decision-making but also addressing disparities in care, particularly for underserved populations.” Her research poster, produced in collaboration with Adelina Ailarov, MS ’23, and Vicky Chen, MS ’23, will be presented at the 36th International Nursing Research Congress this summer.

An immigrant from Haiti, where people commonly lack access to health care, Delma has long been passionate about health care equity. Her interest in reducing disparities—and engaging in research, community service, and advocacy—brought her to Kisumu, Kenya, where she volunteered as a clinical nurse practitioner; Los Angeles and Mexico, where she advocated for immigrant rights; and the global nonprofit Partners in Health, which she joined to advocate for legislation eliminating barriers to care. Now, her passion is driving her in her pursuit of a doctoral degree as an advanced practice nurse.

Through her research, Delma has found that predictive AI models can identify and allow for early interventions in at-risk patients by analyzing subtle patterns in their data. This is especially helpful for underserved populations who often lack access to consistent care, she says. “AI tools like conversational assistants can help patients manage their diabetes by guiding them through complex tasks like insulin adjustments, even when their health literacy or access to specialists is limited,” she says.

Additionally, AI-powered remote monitoring devices can extend high-quality care to rural or resource-limited areas, ensuring that patients receive timely interventions. “These innovations address key barriers, such as gaps in communication and access to resources and biases in care delivery,” Delma says. “AI’s true potential lies in its ability to make health care more efficient, accessible, and equitable for everyone.”

Concerns about Disparities and other Potential Downsides

When she arrived at Columbia, Delma had limited knowledge of generative AI. The more she learned, the more interested she became in the need for regulations and policies to protect data privacy and to avoid algorithm bias that can amplify rather than diminish health care disparities. “If Black patients are hesitant about going to the doctor or being part of certain studies, then they are not providing the type of data that AI’s predictive abilities need to perform well for their community,” she points out. “If there’s bias in our data or health care practice, then AI is going to reproduce that bias,” she adds.

Such bias has skewed skin cancer studies, for example, which historically have lacked data from Black populations and thus produced outcomes disproportionately benefiting white patients, Wright says. “AI learns what you teach it. If you teach it biased information, then it will be a biased machine.”

Bakken underscores the importance of learning how to recognize this pitfall. “Clinical students must know how to assess algorithms and their potential biases,” she says. They also must refrain from viewing AI as a gold standard, she adds. “People need to be thoughtful about AI’s use,” she says. “We can’t accept its findings at face value. AI is a triad of clinical expertise, patients’ lived experience, and what algorithms might tell you about patients who are at risk for certain diseases. It’s just another tool.” What’s more, she adds, “there must be a human component to it.”

No Replacement for Human Touch or Learning

What concerns many about AI is its potential to replace the human component in nursing care. “AI is never going to replace human interaction,” Hickey says. “It can’t sit at the bedside with a patient and hold their hand. It can’t sit next to the parent of a sick child or give them kindness, empathy, and compassion.”

Topol and Delma agree. “AI isn’t taking over who we are and how we approach our practice,” Topol says. Adds Delma, “AI is really meant to augment our role as clinicians but not replace the human aspect of health care. We want to use it to enhance our practice and diagnostic skills but not replace the compassion and empathy that nurses are known to provide.”

Other concerns about AI include patient data breaches and the need for academic guidelines. “As a student, I’ve found AI to be very overwhelming,” Maslag notes, citing his uncertainty about the rules, regulations, and ethics of using it. “Some professors will use AI to create quizzes for us to use in class,” he says, “while others will not allow us to use it for our own studying.”

Delma cautions that “reckless use” of AI could undermine students’ education. “Students could use AI in ways that are not in alignment with their learning, as a quick way to get answers without digesting what’s being taught,” she says. “If schools aren’t ahead in creating AI guidelines or policies of use, then we’re doing a disservice to students,” she says, pointing to personalized study plans that AI can create, and simulated patient encounters that it can facilitate. “Asking AI to correct your grammar is okay but asking it to do the work for you is not doing your due diligence and is setting yourself up for failure,” she says.

A Tool for the Future

In addition, as students learn to navigate AI themselves, they must embrace their responsibility as emerging providers to teach patients how to use it. “Patients are going to use AI by asking Google questions like ‘What is diverticulitis?’ without knowing how to interpret the answers,” Hickey says. “We need to convey to students the importance of making sure patients understand the benefits and limitations of what AI can tell them.”

Indeed, there is much to learn about AI’s positive and negative impacts on students’ experiences in the classroom and, later, at the beside. “It will be important for us as nurses to understand AI so we can help patients understand it—because that’s where the future of health care is heading,” Maslag says. As applications for AI in health care expand, nurses will need to have a say in its use, he says. “The SAC can provide guidance to the school as to the best ways to advance the role of nurses in the implementation of AI in our field,” he observes.

Says Delma, “By prioritizing our voices, the OAI is empowering us to become leaders when it comes to introducing AI into health care, addressing ethical issues like health disparities [and] making sure we’re involved in the development of curriculum, research, and policy, as well as the optimization and management of AI tools,” she says. “As emerging nurses, we must have an impact on the tools that we’ll be using.” 

Illustrations by Benedetto Cristofani; Photographs by Jörg Meyer

This article originally appeared in the Spring 2025 issue of Columbia Nursing Magazine.