Retiree: How to use AI
R.I.S.E.: Talk to AI Like You Mean It
Listen. You've probably tried asking a chatbot something and gotten back a wall of text that missed what you actually needed. Or you asked it to help you understand your medications and it went off the rails into things that had nothing to do with your situation.
That's usually not the AI being dumb. It's you and the AI talking past each other.
There's a framework that fixes this. It's called R.I.S.E., and it's simple enough that I can teach it to you right now, at the kitchen table. Once you know it, you'll get better answers—clearer, safer, more useful.
R — Role: Who Should This AI Be?
Start by telling the AI exactly what hat to wear.
Don't say: "Tell me about my blood pressure medication."
Say: "You are a patient educator who explains medications in plain language. I'm a 72-year-old managing high blood pressure for the first time."
See the difference? The first one is vague. The AI doesn't know if you want chemistry, history, a joke, or actual help. The second one shows the AI who you need it to be.
Good roles for healthcare AI:
"You are a patient coach helping me prepare questions for my doctor's visit."
"You are a health educator explaining symptoms in plain English."
"You are a medication safety checker—tell me what I need to know before I take this."
"You are a calm, reassuring guide helping me understand what my doctor said."
What not to do:
Don't ask it to be a doctor. ("You are a physician diagnosing my symptoms.") It isn't. It can't be. And it will pretend, which is dangerous.
Don't ask it to be something it doesn't understand. ("You are a cardiologist who specializes in my exact condition.") Specificity is good. Fakery is bad.
I — Input: Give It What It Needs
Now tell the AI what information it's working with. The more specific, the better.
Don't say: "I have pain."
Say: "I'm a 74-year-old woman. For three days, I've had a dull ache on the left side of my chest when I climb stairs. It goes away when I rest. I take aspirin and lisinopril for blood pressure. I don't have shortness of breath or sweating. What should I do?"
See? Now the AI knows:
Your age
Where the pain is
When it happens
What makes it better
What you already take
What you don't have
That's gold. That's the difference between a useful answer and a useless one.
Also tell the AI your rules:
"I want to stay home if I safely can."
"I've already talked to my doctor about this, and she said..."
"I need the answer in simple English, no medical jargon."
"Tell me when you're uncertain."
S — Steps: Make It Think Out Loud
Don't ask the AI to jump to the end. Ask it to think through the problem, one step at a time.
For learning:
"Walk me through how to understand this blood test result. First, explain what each number means. Then, tell me which ones are normal for my age. Then, ask me 3 questions to figure out if I should worry."
For decision-making:
"Help me think through whether I should see my doctor today or wait. First, list the warning signs I should never ignore. Second, tell me what signs I have and which ones I don't. Third, help me decide."
For understanding something scary:
"My doctor mentioned 'atrial fibrillation.' Help me understand this by: first explaining what a normal heartbeat does, second explaining what goes wrong, third telling me what questions I should ask my doctor."
Why step-by-step? Because it slows the AI down. It makes it show its thinking instead of just landing on an answer. And it gives you a chance to catch it if it goes wrong.
E — Expression: Say How You Want the Answer
Tell the AI exactly how to format the answer so you can use it.
Bad: "Tell me about managing diabetes."
Good: "Tell me about managing diabetes, formatted as: (1) three things I can do today, (2) a list of foods that are safe, (3) three questions I should ask my doctor at my next visit. Use bullet points and simple language."
Some formats that work well:
Bullet points — when you want quick facts
A table — when you're comparing things (like medication side effects)
Step-by-step numbered list — when you need to do something
A dialogue — when you want to practice what to say to your doctor
Before and after — when you want to see the change clearly
Q&A format — when you have lots of questions
Putting It Together: Three Real Examples
Example 1: You Want to Understand Your Test Results
Role: "You are a patient educator who explains lab results in plain language."
Input: "I'm a 70-year-old man. My doctor gave me these results but I didn't understand the visit. Here's what she said: 'Your A1C is 7.2, your creatinine is 1.1, and your eGFR is 62.' I have diabetes and high blood pressure. I want to know if these are good, bad, or if I should worry."
Steps: "First, explain what each test measures in one sentence. Second, tell me if my numbers are good for someone my age with my conditions. Third, ask me three questions to help me understand what I should do next."
Expression: "Format this as: (1) What Each Number Means, (2) Is This Good or Bad?, (3) Three Questions for Your Doctor. Use bullet points and keep it to one page."
Example 2: You're Preparing for a Doctor's Visit
Role: "You are a healthcare coach helping me get ready for a doctor's appointment."
Input: "I'm seeing my cardiologist about chest discomfort when I exercise. I've had this for two weeks. I'm on metoprolol and lisinopril. I want to know if I should be worried, and I want to ask smart questions so I understand what's going on."
Steps: "First, list the warning signs that mean I need to go to the ER instead of waiting for my appointment. Second, help me organize what I'll tell my doctor (when the pain started, what it feels like, what makes it better). Third, give me five questions I should ask to understand my situation better."
Expression: "Give me a one-page guide I can print and bring with me. Use bullet points. Make it simple enough that I can read it in the waiting room."
Example 3: You Want to Understand a Medication
Role: "You are a medication safety educator."
Input: "My doctor just prescribed me atorvastatin for cholesterol. I'm 72, I take metformin and lisinopril already, and I have mild kidney disease. I want to know: Is it safe with my other medicines? What should I watch for? What questions should I ask?"
Steps: "First, tell me in plain English what this drug does. Second, list any serious interactions with my other medicines or my kidney condition. Third, tell me what side effects I should report to my doctor immediately versus which ones might go away. Fourth, give me five questions to ask my pharmacist."
Expression: "Format as: (1) What This Drug Does, (2) Safety Concerns for Me, (3) Red Flags to Report Now, (4) Questions for My Pharmacist. Keep it to one page, bullet points, no medical jargon."
The Safety Rules (Non-Negotiable)
Every time you use AI for healthcare questions, add these instructions:
Always say this:
"Do not give me a diagnosis or tell me I have a specific disease."
"Your job is to help me understand, not to replace my doctor."
"Tell me clearly when you don't have enough information to help."
"If you see any warning signs in what I've described, point them out so I know to call my doctor immediately."
These aren't suggestions. These are your guardrails. They keep the AI from pretending to be something it's not, and they remind you that a real person—your doctor, your nurse, someone who knows you—is always in charge.
What Gets Better When You Use R.I.S.E.
I've watched people use this framework, and here's what changes:
The AI actually understands what you need. No more walls of text about things you didn't ask.
You stay in control. You're directing the conversation, not following it wherever the AI wants to go.
The answers are useful. Because they're formatted the way your brain works, not the way the AI's default works.
You catch mistakes faster. Because you're reading step-by-step thinking, not just a final answer.
You feel more confident. You're not passively receiving information. You're actively using a tool to think more clearly.
One Last Thing
This framework works because it treats AI like what it actually is: a thinking partner, not an expert.
A good thinking partner doesn't tell you what to do. A good thinking partner asks good questions, stays in their lane, and makes you smarter than you were before.
If you find yourself using an AI that does something different—that pressures you into decisions, that won't admit uncertainty, that talks over you instead of with you—stop. That's not a tool. That's a trap.
Use R.I.S.E. Start small. Practice with something low-stakes—understanding an article, preparing a question, organizing your thoughts. Once you're comfortable, use it for bigger things.
And if you try this and want to tell me what worked? I'm listening.
