Chapter 2 · AI Basics for Healthcare Professionals

“You don’t need to understand the engine to drive the car.”

What Is AI, Really?

In simple terms: AI is software that can understand language, find patterns, and generate useful responses. The AI tools relevant to nurses fall into a few categories:

Large Language Models (LLMs)

Examples: ChatGPT, Claude, Google Gemini
What they do: Read your question, generate human-like text responses
For nurses: Draft notes, summarize research, create educational materials

Clinical Decision Support Systems (CDSS)

Examples: Epic’s AI features, Sepsis Watch, VisualDx
What they do: Analyze patient data and flag potential issues
For nurses: Sepsis alerts, drug interaction warnings, risk scoring

Speech-to-Text / Ambient Listening

Examples: Nuance DAX, Abridge, DeepScribe
What they do: Convert spoken words into structured documentation
For nurses: Chart while talking to patients, dictate notes

How to Talk to AI (Prompting 101)

AI responds to “prompts” — your instructions. Better prompts = better results.

The CARE Framework for Nursing Prompts

Context — Set the scene (your role, the situation)
Ask — State what you need clearly
Restrictions — Set boundaries (HIPAA, scope, format)
Example — Show the format you want

Example: Creating a Patient Education Handout

Bad prompt:

“Tell me about diabetes.”

Good CARE prompt:

“I’m a registered nurse creating a patient education handout. Write a 1-page guide on Type 2 diabetes management for a newly diagnosed patient. Use 6th-grade reading level. Include: diet basics, exercise recommendations, blood sugar monitoring, and when to call the doctor. Do NOT include specific medication names or dosages (I will add those based on the physician’s orders). Format with bullet points and short paragraphs.”

Setting Up Your AI Workspace

Step 1: Choose Your Tool

Tool	Cost	Best For	HIPAA Note
ChatGPT Free	$0	Learning, drafting	NOT HIPAA compliant
ChatGPT Plus	$20/mo	Faster, smarter responses	NOT HIPAA compliant
Claude Free	$0	Nuanced writing, longer documents	NOT HIPAA compliant
Claude Pro	$20/mo	Heavy usage	NOT HIPAA compliant
Facility-approved AI	Varies	Clinical workflows	Check with IT

Important: Use consumer AI tools for template creation, learning, and personal development. Use facility-approved tools for anything involving patient data.

Step 2: Create a Prompt Library

Start a document (Notion, Google Docs, or even a paper notebook) with your most useful prompts. Organize by category:

Documentation prompts
Patient education prompts
Research prompts
Career development prompts

As you work through this book, you’ll build a comprehensive library you can reuse daily.

Step 3: Practice the Feedback Loop

AI rarely gives a perfect answer on the first try. Use this cycle:

Give a prompt → Get a response
Identify what’s good and what’s wrong
Give feedback: “Good structure, but simplify the language” or “Add a section about warning signs”
Repeat until the output meets your standard

What AI Cannot Do in Healthcare

Be clear-eyed about limitations:

Cannot replace physical assessment — AI can’t auscultate lungs, palpate an abdomen, or observe skin changes
Cannot make clinical decisions — AI can suggest, but the nurse decides
Cannot guarantee accuracy — AI “hallucinates” (generates false information with confidence)
Cannot understand context like you do — AI doesn’t know your specific patient, your unit culture, or your facility protocols
Cannot be held accountable — You are responsible for actions based on AI-generated content

Action Items

[ ] Try ChatGPT or Claude: ask it to explain a nursing concept you know well — evaluate its accuracy
[ ] Write your first CARE-framework prompt about a common task on your unit
[ ] Start your Prompt Library document with 3 categories
[ ] Practice the feedback loop: refine one AI response through 3 iterations

Next → Chapter 3: HIPAA, Ethics, and AI Boundaries