María had just finished reviewing a 200-page research draft when her colleague asked a simple question: could a computer have helped? That question led María and her team at the Universitat Jaume I in Castelló, Spain, down a two-year road of careful study — and eventually to a surprising answer.
Generative AI tools, the same chatbots now used by millions of people worldwide, could indeed help nurses do research faster. But using them carelessly could also spread errors, bias, and harm. So María's team did something no one had done before: they wrote down exactly how nurses should use these tools responsibly. Their work, published in the journal Enfermería Clínica, gives researchers ten clear guidelines.
The team came from two groups — the Nursing Research Group, called GIENF-241, at the university, and the eNURSYS group at Fisabio, a public health research center. Both groups study how technology can improve patient care. Their project, known as NURSIA, focuses on information systems, healthcare technology, and measuring whether nursing care is truly high quality.
María and her colleagues found that AI tools can speed up parts of research, like searching scientific papers or drafting questions for studies. But they also found a dangerous habit: people tend to trust AI answers too quickly. The researchers call this the "hallucination" problem — AI can sound completely confident while giving facts that are simply wrong. It might invent study citations or twist the meaning of real research. The solution, they argue, is simple but essential: never trust anything an AI produces without checking it yourself.
This warning matters especially for nursing research, where much of the work involves listening to patients describe their experiences in their own words. AI tools might summarize those conversations in ways that sound reasonable but miss the whole point. A patient's story about feeling scared in a hospital might become just another data point, losing the human details that nurses actually need to understand.
The researchers also pointed out another risk: AI systems often reflect the biases already built into science. Most scientific studies come from wealthy, English-speaking countries and focus on diseases and drugs. That means AI might steer nurses away from the everyday challenges nurses actually face, like how to comfort a frightened child or support a family dealing with long-term illness at home.
So what should nurses actually do? The team's ten guidelines include using AI only when it genuinely helps, never typing private patient information into these tools, always explaining in research papers when and how AI was used, and regularly asking whether the technology is making research better or just faster. The guidelines also remind scientists to follow new rules like Europe's Artificial Intelligence Act, which sets legal standards for how these systems can be used.
María sees this not as a rejection of new technology but as a welcome guide. "We want nurses to use these tools without losing the heart of their work," she explained. "The goal is smarter research, not just faster research." For a profession built on listening and caring, that distinction may be the most important one of all.