The future of nursing education and professional development
- Artificial intelligence (AI) and prompt engineering serves as an innovation within nursing and nursing education.
- Nursing educators and professional development specialists require training to create prompts that deliver meaningful AI responses.
In the ever-evolving healthcare landscape, cutting-edge technologies have become pivotal in shaping the future of nursing education. The intersection of artificial intelligence (AI) and prompt engineering stands as a beacon of innovation; however, nursing educators and professional development specialists require prompt skills to create meaningful and accurate AI responses.
General FAQs about AI
What’s AI? This broad discipline of computer science aims to develop systems capable of performing tasks that traditionally require human intelligence. Types of AI include nongenerative or traditional, which recognizes patterns and makes predictions (for example, Netflix and Amazon), and generative (a subset of traditional, which creates new content based on the information used to train it). Prompt engineering works within limited-memory AI, a component within a generative AI system that allows it to remember and use past interactions to improve new content generation.
How nurses can conquer their fear of AI
Artificial intelligence in nursing
What’s a large language model? A large language model (LLM) is an advanced AI program that recognizes and generates responses using a transformer model neural network. The neural network is trained on extensive data sets from disparate sources, including the internet. The “GPT” in ChatGPT, for example, stands for generative pre-trained transformer. An LLM depends on the quality of these data to effectively learn natural language. It’s then enhanced through deep learning techniques, which allow the LLM to predict and understand how language elements interact. LLMs are refined for tasks such as responding to queries or translating languages via specialized tuning processes.
What’s natural language processing? Natural language processing (NLP), an area of AI, aims to bridge the gap between humans and machines by enabling them to communicate effectively through natural language. NLP uses advanced algorithms and techniques to process, analyze, and understand human language in all its complexity, including grammar, syntax, semantics, and pragmatics. Ultimately, NLP enables machines to comprehend human language in a way that’s useful and meaningful to perform a wide range of tasks, such as language translation, sentiment analysis, speech recognition, and question-answering.
General FAQs about prompt engineering
What’s prompt engineering? Prompt engineering is the deliberate and strategic formulation of instructions or inputs (prompts) given to an AI system to produce desired results. The quality and specificity of the prompt significantly influence the output generated by the AI model. Prompt engineering involves refining the presentation of a question or command. This refinement ensures clarity and specificity to achieve more accurate and relevant responses. This critical approach helps to fine-tune language models and optimize their performance for specific applications, such as healthcare, which has crucial language nuances.
Isn’t prompt engineering the same as performing a Google search? AI responses differ from a typical Google search because users can engage with the AI conversationally, provide feedback, and ask for revisions. A Google search delivers links to information, but a prompt in generative AI interacts directly with a significant language model to generate a response and summarize information from multiple sources automatically. Both necessitate a certain skill level to get the most accurate and relevant results. A Google search requires the right keywords, and generative AI requires the correct prompt.
What are the elements of successful prompt engineering? You can use several techniques to create well-engineered initial prompts for LLMs. First, you must clearly define the desired outcome and purpose of the interaction with the model. This involves specifying the information or response required and identifying potential challenges or nuances related to the subject. A template for creating the initial prompt should include four sections: task, role, audience, context, and instructions. (See Prompt template.)
Prompt template
Using a template, a nurse educator can create an initial prompt that includes a specific action-oriented task, a role (who’s performing the task), an audience (who the educator is addressing), context (relevant information and background to elicit the ideal output), and instructions (task specifics, length, writing tone and style, details about the target audience):
- Task: Craft a nursing care plan.
- Role: From the perspective of an emergency nurse
- Audience: For a newly graduated nurse with a bachelor of science in nursing
- Context: In a Level 3 trauma center emergency department, a 57-year-old male Black patient with hypertension and type 2 diabetes presents with chest pain and shortness of breath.
- Instructions: Consider prioritization of interventions and communication strategies. The care plan should be only 500 words. The tone should be professional, formatted as a checklist with accompanying instructions. Arrange checklist in descending order of importance and criticality. Ensure response is consistent with HIPAA and other data privacy regulations and devoid of bias toward any legally protected classification and that it follows current medical practice.
AI prompts seldom are one and done. However, the more specific the request, the closer the AI’s output will get to meeting your requirements, and the fewer prompt refinements and additional prompts you’ll need. If the prompt generates less than 60% of the output you envisioned, start again by clarifying and refining. Delimiters such as three double quotation marks, @@@, or ### can help the AI process and parse the data correctly within the prompt. For example, you could input the following: “Review the following prompt and make recommendations on how to improve the prompt:…”
You also can prompt the AI to assist you. For example, you could input the following: “I need to generate a nursing care plan. Ask questions one at a time to refine this prompt. Once you have enough information, generate the nursing care plan.”
Benefits of prompt engineering in nursing education
How does prompt engineering contribute to personalized learning experiences for nurses? Prompt engineering tailors educational content to individual needs, preferences, and knowledge levels. Through carefully crafted prompts, AI tools can adapt to different learning styles and address specific learning objectives. For example, you could design a prompt to simulate a real-life patient case, allowing nurses to practice critical thinking and decision-making in scenarios relevant to their specialty. Another example involves adaptive assessments where prompts adjust based on a nurse’s responses, providing a customized evaluation of their skills and knowledge.
This adaptability ensures that nurses receive a more individualized and effective learning experience, which helps enhance their proficiency and competence. By leveraging advanced ideas, innovation, and knowledge, prompt engineering can contribute to more effective use of technology in healthcare settings, including the quality of information retrieval, decision support, and educational tools.
How does prompt engineering address issues of bias and equity in nursing education? Prompt engineering is emerging as a crucial tool in addressing bias and promoting equity in nursing education. One proactive approach involves careful crafting of prompts to ensure inclusivity and mitigate biases in AI systems. Use inclusive language and diverse representations to avoid stereotypes and to reflect the diversity of patients and healthcare professionals. Integrated through prompt engineering, bias-detection mechanisms can aid identifying and correcting biases in educational content and assessments. Prompts that encompass various backgrounds support cultural competence and prepare nurses to provide equitable care. Addressing social determinants of health and ensuring transparency in algorithmic decision-making also contribute to equity.
This effort requires collaboration with diverse stakeholders and ongoing monitoring, evaluation, and bias mitigation training for educators. Using these strategies, prompt engineering becomes a powerful ally in creating an inclusive learning environment and preparing nursing students to navigate healthcare with sensitivity and fairness.
What are some real-world examples of prompt engineering being applied to nursing education scenarios? Examples include tailoring educational content based on an individual learner’s needs, AI-powered simulated patient interactions for practicing communication and decision-making skills, and realistic clinical scenarios for simulation exercises. (See Prompts and real-world nursing.)
Prompts and real-world nursing
Consider the following examples of prompts that can elicit responses relevant to real-world nursing education scenarios:
-
- Medication administration practice: You are the nurse responsible for administering medications to a patient with multiple chronic conditions. Describe the steps you would take to ensure safe and accurate medication administration, including dosage calculations and patient education.
- Patient communication in palliative care: You are a nurse providing care to a terminally ill patient in a palliative care setting. Outline a communication plan that addresses the patient’s physical and emotional needs, involving both the patient and their family.
- Interprofessional collaboration: You are a nurse and part of an interprofessional team caring for a patient with complex healthcare needs. Describe how you would collaborate with other healthcare professionals, ensuring seamless communication and coordination of care for a patient in a critical care unit.
- Cultural competence in nursing practice: You are the nurse assigned to care for a patient from a different cultural background. Describe how you would provide culturally competent care, taking into consideration cultural beliefs, preferences, and potential communication challenges.
Managing the limitations of prompt engineering
What are the primary limitations associated with prompt engineering? Prompt engineering in NLP has its limitations, including the potential to overfit, which occurs when an algorithm becomes too focused on the training data. Consequently, the algorithm fails to generalize new and unseen data. In addition, relying on well-crafted prompts can introduce a dependency on prompt quality, which can lead to inaccurate model outputs in response to poorly designed prompts.
A risk for bias and subjectivity also exists as the framing of prompts may influence the model’s responses. Prompt-based approaches may require help with creativity and adaptation to new tasks. This increased human involvement in prompt creation can prove time-consuming. Semantic drift (how meanings of words evolve) over time may cause prompts to lose alignment with intended tasks, and handling ambiguity proves challenging for prompt-engineered models. In addition, the limited explainability of model outputs poses difficulty in understanding the reasoning behind responses.
Users who rely solely on a model’s conclusions without understanding how the model arrived at them are more likely to accept them without question. This can amplify any pre-existing biases in the data used to train the model or introduce new biases through the model’s decision-making mechanisms. For example, a hospital may use an AI-based diagnostic decision support system to help diagnose sepsis. However, when the system diagnoses sepsis, it doesn’t explain how or why it reached that conclusion. Biased recommendations from the AI model may lead to disparities in patient care, such as unnecessary treatments or interventions.
How can I mitigate the challenges of prompt engineering in developing effective language models? Understanding and leveraging the strengths of the language model can help mitigate challenges. One way to avoid misinterpretation involves writing specific prompts. Providing context and prompt examples also can help improve the model’s understanding and performance. Experimenting with different prompts and fine-tuning the model based on the results is essential.
Fair and accurate model outputs rely on diverse and unbiased prompts, human evaluation, and adversarial training (deliberately feeding false data into the model to cause errors). Adversarial training helps the model learn to recognize false inputs and improve output accuracy. For instance, if you train a model to recognize images of dogs, you’ll input many dog images. However, the model may sometimes get confused and output images of a large cat or some other animal. By building the model using images of other animals along with an indication that these images aren’t dogs, the model learns to identify dogs more accurately.
Before disseminating or using the output generated by the language model, run it through a human review process to help identify errors, biases, or inappropriate content. The content should meet the highest quality standards. Prompt engineering is an iterative process that requires continuous refinement; always consider the potential for defects, biases, and ethical concerns.
What specific challenges or considerations does prompt engineering present in nursing education? Prompt engineering in nursing education faces some challenges, including navigating the complex language used in medical contexts. Crafting effective prompts requires educators to have a deep understanding of evolving healthcare practices and terminology, as well as the ability to challenge learners. Collaboration among healthcare experts, educators, and AI engineers can help ensure the creation of accurate and educationally effective prompts.
Other challenges include technical issues, standardization of learning curricula, and potential lack of clinical judgment development. Strike a balance between AI tools and traditional methods to mitigate these risks and ensure well-rounded nursing education.
What risks are associated with over-reliance on prompt-engineered AI in nursing education? Prompt-engineered AI requires careful consideration of risks to ensure a balanced integration of technology and traditional methods. For example, overdependence on predefined prompts may discourage independent analysis and problem-solving, which may lead to the loss of critical thinking skills among nursing educators. In addition, an AI tool might cover only some of the challenges nurses may encounter in their careers, limiting their exposure to diverse scenarios.
Ethical considerations related to academic work also present a challenge. Generative AI use and prompt engineering in education is complex and requires addressing varied learning styles, providing quality feedback, integrating with clinical practice, and maintaining learner engagement. For example, nursing students who use generative AI and prompt engineering must ensure that their academic work accurately reflects their learning journey and commitment to ethical practice in healthcare. A collaborative and thoughtful approach can help overcome these challenges and ensure the development of ethical and practical AI tools for nursing and healthcare training.
Implementation and integration
How can nursing educators incorporate prompt engineering into their curriculum effectively? The integration of prompt engineering into the curriculum requires a strategic approach. First, you must receive training in prompt engineering, including prompt development, AI interpretation, and classroom integration strategies. Collaboration with AI experts can help you understand the technology’s capabilities and limitations. Identify specific learning opportunities where prompt-engineered AI can enhance the educational experience, such as simulated patient interactions or adaptive assessments.
Clearly defining educational objectives (critical thinking, clinical decision-making, personalized learning) can help you develop prompts. In addition, you should actively solicit learner feedback to refine prompts and regularly assess learning outcomes to ensure alignment with educational objectives. By taking a strategic and collaborative approach, you can effectively integrate prompt engineering into your curriculum, offering learners a transformative learning experience that aligns with the evolving landscape of healthcare education.
Do specific training requirements exist for nursing educators and professional development specialists to understand and use prompt-engineered AI? Effective use of prompt engineering requires faculty familiarity with the core concepts of AI. This includes understanding how to design prompts, tailor queries to achieve desired outcomes, and formulate instructions for AI systems. Practical exercises and relevant examples, specifically in nursing scenarios, can enhance prompt creation skills. Seek training on key issues, such as patient privacy, data security, bias mitigation, equity, and responsible use of AI tools.
In addition, learn how to effectively combine AI tools with traditional teaching methods for designing cohesive and integrated learning experiences that enhance specific learning objectives. This includes guiding students in how to analyze AI-generated responses, assess the relevance of provided information, and derive meaningful insights. To facilitate discussions around AI-generated content, encourage student reflection and collaboration with AI experts. Students also should have access to resources and support that foster a collaborative environment where they can seek guidance, share insights, and collectively enhance their proficiency with AI tools.
Future trends and innovations
What are the emerging trends in prompt engineering and considerations for the future development of prompt engineered AI tools in nursing? Trends within nursing education include a continued move toward personalized and adaptable prompts, which can help tailor learning experiences to individual preferences and knowledge gaps. Integrating multimodal prompting and incorporating text, images, and voice can help enrich the learning experience. Realistic simulations will likely become more immersive, allowing nurses to practice decision-making in lifelike scenarios within a virtual environment. Explainable AI (ability of AI models to provide humans with decision-making explanations) addresses the demand for transparency by clarifying reasoning and decision-making processes. Collaborative learning prompts also may become more prevalent, emphasizing teamwork and shared problem-solving among nurses.
How should an educator research the factual basis (relevancy) of a student’s submission if they’re concerned about over-reliance on generative AI? Academic institutions expect educators to hold students accountable and to ensure graduates are competent, professional, ethical, and risk-averse. Although no method or tool can guarantee 100% accuracy in detecting AI-generated content, educators can consider a few approaches. Online tools analyze text characteristics, such as sentence structure, word choice, and predictability to detect AI-generated content. Plagiarism checkers primarily detect copied content, but some now include AI-generated text detection. Also consider the student’s work quality. Sudden improvement in a student’s work may indicate AI assistance, especially if they previously struggled with complex concepts or writing. Comparing a student’s previous work with a current submission can help identify any significant changes in writing style or vocabulary that may suggest AI use.
Students use AI tools, and educators must take necessary measures to ensure academic integrity. To foster an environment where students feel responsible for their learning, encourage collaborative engagement and discussion about academic integrity and the ethical use of AI.
Strike a balance
Integrating prompt-engineered AI tools in nursing education presents a transformative landscape with promise and potential pitfalls. Although these tools can offer personalized learning experiences, adaptive feedback, and innovative educational approaches, the risk of diminishing critical thinking skills, limited exposure to diverse scenarios, and reduced development of interpersonal skills highlight the importance of maintaining a balanced educational approach. Ethical concerns, potential technology dependence, and the standardization of learning further underscore the need for thoughtful integration of AI tools into the nursing curriculum.
Striking this balance requires continuous collaboration among educators, AI experts, and clinicians. By acknowledging and actively addressing these challenges, nursing education can harness the benefits of prompt-engineered AI while also ensuring the development of well-rounded, competent, and ethically aware healthcare professionals.
Editor’s note: The authors did not use an OpenAI source to create prompts included in this article, but they did use OpenAI sources—Copilot (bit.ly/4dcz7Tx) and ChatGPT 3.5 (chat.openai.com/chat)—to validate responses.
The authors work at the American Nurses Association. Jennifer Shepherd is director of nursing education and product management. Donald Griesheimer is senior education technology designer.
American Nurse Journal. 2024; 19(6). Doi: 10.51256/ANJ062414
References
Acar OA. Are your students ready for AI? Harvard Business Publishing. June 15, 2023. hbsp.harvard.edu/inspiring-minds/are-your-students-ready-for-ai
Bahrini A, Khamoshifar M, Abbasimehr H, et al. ChatGPT: Applications, opportunities, and threats. IEEE Systems and Information Engineering Design Symposium. 2023. 2304.09103.pdf (arxiv.org)
Brown S. Machine learning, explained. MIT Management Sloan School. April 21, 2021. https://mitsloan.mit.edu/ideas-made-to-matter/machine-learning-explained
Charow R, Jeyakumar T, Younus S, et al. Artificial intelligence education programs for health care professionals: Scoping review. JMIR Med Educ. 2021;7(4):e31043. doi:10.2196/31043
Cloudflare. What is a large language model (LLM)? cloudflare.com/learning/ai/what-is-large-language-model
DeepLearning AI. Natural language processing. deeplearning.ai/resources/natural-language-processing
Fox N. How Google organizes information to find what you are looking for. Google. December 3, 2020. blog.google/products/search/how-google-organizes-information
Heston TF, Khun C. Prompt engineering in medical education. Int Med Educ. 2023;2(3):198-205. doi:10.3390/ime2030019
IBM. What is machine learning (ML)? ibm.com/topics/machine-learning
IBM. What is prompt engineering? ibm.com/topics/prompt-engineering
IBM. Understanding the different types of artificial intelligence. October 12, 2023. ibm.com/blog/understanding-the-different-types-of-artificial-intelligence
Microsoft Build. Introduction to prompt engineering. March 29, 2024. learn.microsoft.com/en-us/azure/ai-services/openai/concepts/prompt-engineering
O’Connor S, Peltonen LM, Topaz M, et al. Prompt engineering when using generative AI in nursing education. Nurse Educ Pract. 2023;74(2). doi:10.1016/j.nepr.2023.103825
OpenAI. Teaching with AI. openai.com/blog/teaching-with-ai
Pandian S. Explainable AI (XAI) for AI and ML engineers. November 15, 2022. analyticsvidhya.com/blog/2022/10/explainable-artificial-intelligence-xai-for-ai-ml-engineers/
Shepherd J. Unlocking the future of nursing education and continuing professional development by embracing generative artificial intelligence and advanced language models. Interdiscipl J Partnership Stud. 2023;10(2). doi:10.24926/ijps.v10i2.5614
Toner H. What are generative AI, large language models, and foundation models? Center for Security and Emerging Technology. May 12, 2023. bit.ly/3YcPVCD
Key words: artificial intelligence, AI, prompt engineering, natural language processing, nursing education
