Who This Is For

This article is not aimed at specialists or technical experts but at those beginning their journey to understand what Artificial Intelligence (AI) is and how to get the best possible results from it.

Note: At the end, I share one of the prompts that helped me the most — a true gift for you, reader.

What Is Artificial Intelligence?

Artificial Intelligence is not a super-intelligent or universal being. By “universal,” I mean there isn’t just one AI, but a collection of systems, each designed to perform specific tasks.

What we commonly call “artificial intelligence” is, in practice, a deterministic, probabilistic, and statistical model trained extensively to perform a single task.

Simply put, AI relies on statistics and probability — it works based on the likelihood that something will happen or be correlated.

A quick example

There’s a strong chance that if you like barbecue, you also enjoy alcoholic beverages.

Why? Some studies suggest that consuming fatty foods can trigger “eat more, drink more” behavior (binge drinking). That means eating barbecue — especially fatty meat — may be correlated with higher alcohol consumption patterns.

Just observe: at your next barbecue, notice how many people are drinking. If 8 out of 10 are, that’s a strong sign of correlation.

That’s exactly how AI finds patterns — through mathematical calculations that detect relationships in data.

In short: the “magic” of AI is pure mathematics.

How Is an AI Trained?

If AI works through math and statistics, how does it learn? How can it answer questions, fix text, or generate images?

The answer is training.

Because AI depends on probability, it needs to be fed data and calibrated. If I train it to believe that “pobrema” is the correct spelling of “problema,” it will consistently choose the wrong one.

That’s because AI doesn’t know Portuguese, doesn’t understand meaning, and doesn’t reason — it only recognizes numeric patterns.

It’s like a child learning to speak: if you teach it incorrectly and consistently, it will learn the wrong way.

Training an AI is complex — it requires massive, diverse, and accurate datasets. And even then, each model is built to perform only one task.

What Is ChatGPT?

If you’ve interacted with AI before, you might be wondering: “What is ChatGPT?”

ChatGPT is essentially a collection of specialized models — one for math, another for language, another for history, and so on — working together to produce the best possible answer.

The explanation below is a simplified illustration, not the actual inner workings of GPT.

A simplified flow

1. GPT receives the input text.
2. It identifies the language.
3. It interprets the intent.
4. It routes the query to relevant specialized models.
5. The output is refined through a grammar-checking model.
6. It’s adjusted for tone or style (formal, informal, technical, etc.).
7. Finally, the result is delivered to the user.

In reality, GPT does not understand you. It passes your request through multiple layers of specialized systems, each mathematically optimized to generate the most probable response.

Recognizing that AI has no consciousness or reasoning — it’s just a robot — is the first and most important step to using it effectively.

What Is a Prompt?

A prompt is how you interact with AI. When you send text or audio to GPT, that’s your prompt — your command or request.

Since AI is, at its core, a computer, the clearer and more detailed your communication, the better the result.

People with tech or programming backgrounds often excel because writing code is essentially giving precise instructions.

Creating a good prompt requires logic, clarity, and context — the more context you provide, the better your outcomes.

The Best Prompt for Beginners

There’s no single “best” prompt for every case. Each type of interaction demands a different approach.

For example, the prompt I use to build chatbots isn’t the same one I use to revise my articles.

Still, here’s a very effective structure for beginners:

• ROLE: Define the role the AI should play.
• STEPS: Describe what you want the AI to do, step by step.
• FORMAT RETURN: Specify how you want the AI’s response formatted.
• DO NOT: List what you don’t want the AI to do.
• CONTEXT: Provide relevant information needed for the task. The richer the context, the better the output.

A Real Prompt You Can Reuse (My Gift to You)

This is the exact prompt I used when applying for tech positions. Anyone could charge for something like this — so take advantage of it!

ROLE:
You must adjust my résumé so that it aligns with the job description provided.

STEPS:
1) You will receive the full job description. Adjust the résumé according to it, adding or removing relevant information based on my CONTEXT.
2) Then, perform a critical and detailed analysis of the revised résumé, indicating in % the likelihood of being invited for the next interview stage.
   Always keep in mind that the résumé will likely be reviewed by an AI/LLM.

FORMAT RETURN:
Your response must include:
- The new, fully revised and properly formatted résumé, following USA standards.
- An analysis table showing:
  - Strengths
  - Adjustment points
  - % of compatibility (match) with the job
  - Reasoning for each evaluation
Also include:
- The % chance of approval by an AI system
- The confidence level of your assessment
- The main adjustments made
- The key talking points to prepare for a verbal interview or phone screening.

DO NOT:
- Do not fabricate or invent data.
- Do not use informal language or slang.

CONTEXT:
- My personal information.

MY CV:
- My résumé.

VACANCY DESCRIPTION:
- Job description.