How to Prompt AI: The 5-Part Formula That Stops Hallucinations
- D. A. Schippers
- Mar 18
- 7 min read
Updated: Apr 6
The Real Reason Your AI Keeps Getting It Wrong — And the Formula That Fixes It
By Dr. Dave Schippers, Sc.D., CISSP | Founder, Iron Dog LLC
Most people using AI (ChatGPT, Claude, Gemini, or Copilot) today are running it in its weakest mode — and they don't know it.
They type a vague question. They get a vague answer. They decide AI is overrated, unreliable, or just not for them. That conclusion is wrong — but the frustration is completely legitimate. The tool didn't fail. The interface did.
AI is not a search engine. It is not a magic answer machine. It is a direction-following system. What you get out of it is almost entirely determined by the quality of instructions you put in. When instructions are thin, the AI doesn't stop and say it needs more information. It fills the gaps. It guesses. It generates something that sounds right — and very often isn't.
That's not a bug. It's how the system is designed. It's also completely correctable once you understand what's actually happening under the hood.
Why AI Hallucinates — And Why You're Probably Making It Worse
The term hallucination makes it sound like a glitch. It isn't. It's the predictable output of a system that was built to keep responding, even when it doesn't have enough information to respond accurately.
AI models are prediction engines. They predict what a useful answer should look like based on patterns in their training. If your prompt leaves gaps, the model fills them — with confidence, professional language, and complete fabrication. A statistic that doesn't exist. A study that was never conducted. A source that has never been published. All of it delivered in the same calm, authoritative tone as accurate information.
Poor prompts make this dramatically worse in three predictable ways.
Vague prompts force the AI to substitute assumptions for facts. The more undefined your request, the more the model has to invent to complete it.
Requests for facts without asking for sourcing give the AI permission to produce fact-shaped sentences with no obligation to verify them.
Instructions that demand confidence push the model to sound certain even when the answer is genuinely uncertain.
Clear, structured prompts solve this. When you define the role, the goal, the context, the inputs, and the output format, the AI has fewer gaps to fill and less reason to invent. You shift it from performance mode — sounding good — into work mode — being accurate.
That shift is the entire game.
The 5-Part Formula
Every effective AI prompt contains five parts: Role, Goal, Context, Inputs, and Output Format. This framework — developed from years of AI deployment, academic leadership, and organizational systems research — eliminates the ambiguity that causes AI hallucinations. Every strong prompt has all five. Every weak prompt is missing at least one. This 5-Part formula works across ChatGPT, Claude, Gemini, Copilot, and any instruction-following AI model.
PART | WHAT IT DOES |
ROLE | Tell AI who it is. A strategy advisor thinks differently than a search engine. |
GOAL | Define success before you start. What does a useful answer actually look like? |
CONTEXT | Audience, tone, constraints, time frame. The AI cannot guess what you already know. |
INPUTS | Give it your material. Notes, drafts, data. The more you provide, the less it invents. |
OUTPUT FORMAT | Specify the result type. Report, checklist, email, table. Format shapes usefulness. |
This Is a Cognitive Skill, Not a Technical One
Prompting is not programming. It does not require a technical background, a certification, or fluency in any language other than your own. It requires one thing: the discipline to define what you actually need before you ask for it.
That discipline — clarifying goal structure before execution, identifying constraints before output — is the same cognitive habit that separates strong analysts from weak ones, effective leaders from reactive ones, and professionals who use AI as a force multiplier from those who use it as a slightly faster search bar.
The individuals who get the most out of AI are not the ones with the most technical knowledge. They are the ones who think clearly about what they need and communicate that clearly in their prompts. Existing skills — domain expertise, analytical depth, pattern recognition, professional judgment — become advantages the moment they are applied to the prompting process.
What This Changes
Once you apply this 5-Part formula consistently, the tool stops feeling unreliable and starts functioning like a capable professional who understands your field, your constraints, and your standards — because you've told it all three.
A manager prepares for a difficult conversation by prompting AI to model responses and anticipate objections — specific to the actual situation, not a generic scenario.
A researcher moves from raw notes to a structured analysis framework without spending three hours staring at a blank outline.
An executive asks AI to stress-test a strategic decision against second and third order consequences — and gets a response that matches the complexity of the actual problem.
None of this requires advanced AI knowledge. It requires a structured prompt.
Bad - Ineffective Prompt for AI
Plan a trip to Europe for my wife and I. We want to only spend $8,000 to $10,000.
5-Part Formula Super Effective Prompt - Trip Planning Example
ROLE:Act as an experienced luxury travel consultant with deep expertise in European cultural destinations, advising a couple in their late 50s who prioritize comfort, authentic local experiences, and avoiding tourist traps.
OBJECTIVE:Create a detailed 10-day itinerary for Italy — split between Rome and Tuscany — that balances cultural depth with physical pacing appropriate for travelers who prefer quality over quantity.
CONTEXT:Two travelers, ages 57 and 59. Both are experienced international travelers who have been to Western Europe before but not Italy. We enjoy art museums, local food markets, wine, and slow mornings. We do not enjoy large tour groups, theme parks, or packed schedules with back-to-back activities. We walk well but prefer not to stand for hours. We fly into Rome and out of Florence. We have already visited Paris, Amsterdam, and Lisbon.
CONSTRAINTS:$8,000–$10,000 USD total excluding flights. Accommodation style: boutique hotels or small family-run properties — no large chain hotels. Must include: at least one cooking class and one private wine tasting. No more than 3 major tourist sites per day. Travel between cities by train only — no rental cars. Must allow for one full rest day mid-trip.
ANALYSIS REQUIREMENTS:Day-by-day breakdown with morning, afternoon, and evening recommendations • Specific restaurant recommendations for at least 5 dinners — local trattorias, not tourist-facing establishments • Advance booking requirements and lead times for each activity and site • Estimated cost breakdown by category: accommodation, food, activities, transport • One alternative activity for each major planned event in case of closures or weather • Practical logistics: train routes, approximate travel times, which neighborhoods to stay in and why
OUTPUT FORMAT: Structure the response as a day-by-day travel plan with the following
sections:
1. Trip Overview and Travel Philosophy
2. Pre-Trip Booking Checklist
3. Day-by-Day Itinerary (each day with morning, afternoon, and evening)
4. Dining Guide
5. Budget Estimate by Category
6. Practical Logistics
I have shared my prompts with multiple speaking groups. Multiple people have reached out afterwards saying this 5-Part formula consistently changes the ball game. My medical doctor gave up on ChatGPT and found it useless and cumbersome. He tried my 5-Part formula and his comment was "It was spot on. This is a game changer for me. I tested its medical knowledge and was impressed at the knowledge and level of questions and interactions it applied after the prompt."
Want more prompt examples? The 5-Part formula with copy and paste examples is a click away.
FAQ - Frequently Asked Questions
What is the best way to prompt AI? The 5-Part Formula - Every effective AI prompt contains five parts: Role, Goal, Context, Inputs, and Output Format. This framework — developed from years of AI deployment, academic leadership, and organizational systems research — eliminates the ambiguity that causes AI hallucinations.
Why does AI hallucinate? Vague prompts force the AI to substitute assumptions for facts. The more undefined your request, the more the model has to invent to complete it.
Do I need technical skills to use AI effectively? Prompting is not programming. It does not require a technical background, a certification, or fluency in any language other than your own. It requires one thing: the discipline to define what you actually need before you ask for it.
What is a prompt engineering framework? A prompt engineering framework is a template used to provide AI key information to avoid hallucinations. The 5-part formula is an extremely effective framework.
How do I stop AI from making things up? Clear, structured prompts solve this. When you define the role, the goal, the context, the inputs, and the output format, the AI has fewer gaps to fill and less reason to invent. You shift it from performance mode — sounding good — into work mode — being accurate. The 5-Part formula is the best example to use.
About the Author
Dr. Dave Schippers, Sc.D., CISSP is the founder of Iron Dog LLC, and creator of the AmplifAId Intelligence framework — a cognitive architecture model for professionals navigating AI-integrated environments. He is the author of Kill the Leadership Theater and the forthcoming AI for Analogs. His work sits at the convergence of AI strategy, organizational psychology, cybersecurity, and academic leadership.
Through Iron Dog LLC's AmplifAId Intelligence division, Dr. Schippers develops AI twin products, frameworks, and digital content for professionals who want to use AI as a genuine cognitive extension — not a replacement for their own judgment.
Comments