Generative artificial intelligence (GenAI), particularly large language models (LLMs) like ChatGPT and others, is here to stay and having an impact on the CME industry. In a recent session led by Andy Crim and Brian McGowan during the CMEpalooza fall 2023 conference, more than 75% of participants indicated they had used GenAI in their workplace. Thirty-eight percent indicated they had a positive experience and it was very useful, while 38% indicated that their experience has been mixed, with some positive and some negative results. With so many in our industry using these AI models, how can we enhance their use and improve the results? It starts with effect prompt creation.

Prompt Engineering

GenAI language models are trained on vast datasets, enabling them to generate, translate, answer and elaborate on text-based inquiries, or prompts. However, their performance is intrinsically tied to the quality and specificity of the prompts they receive. A well-constructed prompt can lead to detailed, accurate, and contextually relevant responses, while a vague or poorly structured prompt will yield more generalized and less useful results. Effective prompt engineering is crucial to get the most out of your work with GenAI models.

Crafting the Perfect Prompt

The key to crafting an effective prompt lies in clarity, specificity and understanding the GenAI model's capabilities and limitations. Prompts can contain questions, instructions, contextual information or examples. During the CMEpalooza session, Crim and McGowan discussed their experience over the last 11 months exploring best practices in prompt engineering. They shared a list of top 10 prompt best practices compiled initially by asking four of the most popular GenAI models (ChatGPT, Claude, Perplexity and Bard). You can download the tool here.

Top 10 Best Practices in GenAI Prompt Engineering

1. Define the “role” of the response.

• Specify the expertise or perspective from which the response should be anchored.

Example: "You are an expert in adult learning theory, please summarize the theory of andragogy."

2. Define the target audience for the response.

• Specify who will be consuming the output of the prompt.

Example: "Write a summary of the current clinical guidelines in triple negative breast cancer for a non-clinical audience. Assume their topic literacy is that of a high school student."

3. Describe the tone and style of the response.

• Similar to the role and target audience, GenAI has widely varied ways of formulating a response from highly technical to whimsical, and from prose to haiku.

Example: Append your prompt with the following statement, "Your response should be written in casual and familiar tone with a slightly humorous style."

4. Describe the format of the response.

• Be explicit in what form you would like the response as most GenAI models can produce text, code, tables, data visualization, etc. Providing examples of desired output can be helpful.

Example: "Create a table of the 10 most cited research papers on learner engagement. The table should have six columns, and headers should be ‘Article Title’, ‘Author Names’, ‘Journal Name’, ‘Year of Publication’, ‘Number of Citations’ and ‘Summary’. For the ‘Summary’ column, please add a brief summary of the publication in 100 words or less."

5. If you need a specific response, use specific language.

• Utilize precise language to guide the model toward the desired response.

Example: "Explain the clinical challenges with cancer treatment for a newly diagnosed patient with cancer", instead of "What is the worst part about a cancer diagnosis?"

6. Use examples.

• If you can, provide examples of the type of output you want the AI to generate. This can help the AI understand your vision and generate more accurate results.

Example: “Using the question-writing best practices from National Board of Medical Examiners, create five assessment question with feedback from the following lecture transcript.”

7. If you need a creative/brainstorming response, use open-ended language.

• Being too specific will limit the creativity of the response. For creative work, start with more general prompts.

Example: "What is the worst part about a cancer diagnosis?" instead of "Explain the clinical challenges with cancer treatment for a newly diagnosed patient with cancer."

8. Employ chained prompting.

• Modify and build upon prompts iteratively to home in on the desired response.

Example: Start with a broad prompt and then ask follow-up questions to gather specific details.

9. Use the GenAI model as a prompt guide.

• One of most creative and effective prompt strategies is to ask the GenAI to reflect on your prompt before it begins to generate a response.

Example: "Before beginning, what other information do you need to optimize your response to the following prompt? [WRITE YOUR PROMPT]?”

10. Be safe.

• When using any of the large AI models, be highly cautious about sharing confidential or legally privileged information. Many of these models are designed to use your prompts as additional inputs training the AI models. Check the terms and conditions, and check with your leadership and legal teams.

Conclusion

LLMs are conversational models, and better results will result from conversing with them rather than just asking questions. Constructing effective prompts for GenAI is about engaging in a dynamic dialogue where clarity, context and creativity play pivotal roles. By mastering prompt engineering, we can unlock new dimensions of interaction and discovery with GenAI.

Whether you're using GenAI in your professional lives to write portions of a grant request for a CME activity or using it to help your kids create an effective marketing campaign for a school fundraising program, developing this skill can dramatically enhance your AI experience and lead to more insightful, accurate and actionable responses. Navigating and shaping this AI-augmented world requires embracing the art of prompt crafting as a key tool in our digital literacy arsenal.

---

An experienced industry leader with demonstrated success developing businesses in the healthcare/medical education industry over several decades, Kenny Cox, CHCP, FACEhp, utilizes skills in management/leadership, education, marketing, technology, finance, business development and strategic planning to create new opportunities in medical education that lead to improved patient care and, ultimately, enhanced patient outcomes. He has held positions in numerous companies and currently is the VP of CME at ArcheMedX. Kenny has also served on various boards/committees for the Alliance and currently serves on the Almanac Editorial Board.

Andrew Crim, MEd, CHCP, FACEhp, is the director of education and professional development for the American College of Osteopathic Obstetricians and Gynecologists (ACOOG). He has more than 27 years’ experience in adult learning and instructional design of continuing education for health professionals, especially for physicians, nurses and pharmacists. He has developed and overseen thousands of continuing education activities for healthcare professionals in North Texas, throughout the state and around the country. His efforts and interests are focused on using education as a catalyst for clinical improvement and increased safety in healthcare.

Andy serves on the board of directors of the Alliance for Education in the Health Professions and serves on the accreditation subcommittee of the Texas Medical Association. He has received two gubernatorial appointments and continues to serve on the Texas Council for Developmental Disabilities.

Mr. Crim is a Fellow of the Alliance for Education in the Health Professions, and his work has been recognized through numerous professional awards, including the ACEhp’s Felch Award for Outstanding Research in CE and the Award for Innovation in the CPD Enterprise. He was also co-author of a peer-reviewed manuscript based on his research.

Dr. McGowan has served in leadership positions in numerous providers and commercial supporters. He is a Fellow of the Alliance (FACEhp), has served on numerous committees, launched the Outcomes Standardization Project, and hosted the Alliance Podcast Legends Interview series. In 2012, he co-founded ArcheMedX, Inc, a healthcare informatics and e-learning company to apply his research in practice. You can follow him on X (formerly Twitter) (@BrianSMcGowan) and connect with him on LinkedIn.