DRD 55: How High Performers Actually Use Artificial Intelligence Every Day

Most people are using artificial intelligence completely wrong.

In my last post, we broke down the basics of "Prompt Engineering." If you missed that, go back and read it here. But knowing how to talk is only step one. Today, we’re going to talk about who you are talking to.

Because right now, you’re treating your AI like a glorified Google search. You ask a question, it spits out an answer, and you move on.

That’s a waste.

You have the most powerful consultant in history sitting in your pocket, and you’re using it to write generic emails. Let’s change that. Today, I’m going to show you how to turn that chatbot into a thinking partner, a debate opponent, and a decision-making engine.

The "10 Colleagues" Theory

Here is the thing about artificial intelligence models: they are trained on massive sets of data. They are evolving literally as you read this sentence.

But here is the kicker: Consistency is not their strong suit.

If 10 of your colleagues sit in a room, open the exact same AI tool, and type in the exact same prompt at the exact same time... guess what? You will likely get 10 different results.

Why? Because these tools work on probability, not rigid logic. They are predicting the next word, not looking up a static fact in a library.

This variability scares some people. But for us? It’s an opportunity. It means the output isn't fixed—it’s malleable. It means you can shape it.

Stop Starting from Scratch: Train Your AI (Artificial Intelligence)

Did you know you can train your AI?

Most modern tools (like ChatGPT or Gemini) now have a Memory feature or "Custom Instructions." This is a game-changer.

If you don't use this, every time you open a new chat, the AI has amnesia. It doesn’t know who you are, what you do, or what you care about. You have to explain context all over again. Exhausting.

Instead, you need to feed it context so it "remembers" you across conversations.

How to Train It (Without Being Creepy)

You can tell the AI specific things about your professional life so it stops giving you generic advice.

For example, I could tell my AI:

"I am Dr. Arun, a specialist in Transfusion Medicine. I run a blog called ThirdThinker about productivity and leadership. I prefer clear, simple language. My motto is 'No patient shall die in need of blood'."

Once it knows this, every answer it gives me will be filtered through that lens. If I ask for a productivity tip, it won't give me advice for a college student; it will give me advice suitable for a medical professional and leader.

However, a massive warning here: The Safety Stop: Set Your Restrictions

Do not feed it your personal or confidential information.

Unless you are okay with that data potentially being used for training future models, keep it generic.

• Good to share: Public professional info, your writing style, general preferences, your public LinkedIn bio.

• Bad to share: Patient data, financial passwords, private family details, sensitive work secrets.

Be mindful. You want a helpful assistant, not a privacy leak.

How to Set Boundaries and Restrictions (Most People Skip This)

Before using AI seriously, set constraints.

Restrictions improve output quality.

You can say:

• “Do not use buzzwords.”

• “Challenge my assumptions.”

• “If my idea is weak, say so directly.”

• “Use Indian context where relevant.”

• “Avoid Western corporate clichés.”

This prevents shallow, generic answers.

Think of restrictions as guardrails for better thinking.

The Secret Weapon: The PROMPT Framework

Now, let's get to the meat of it. How do you actually get brilliant ideas out of this thing?

You stop asking one-sentence questions. You start building structures.

To get the best results, use the PROMPT Framework. This is a mental checklist to ensure you aren't leaving the AI to guess what you want.

• Persona: Who should the AI act as? (e.g., "Act as a Senior Strategist")

• Request: What specifically do you need? (e.g., "Critique my plan")

• Output: How do you want the answer to look? (e.g., "A table" or "Bullet points")

• Modifiers: Any constraints? (e.g., "Keep it under 200 words", "Be harsh")

• Purpose: Why are we doing this? (e.g., "To pitch to an investor")

• Tone: What is the vibe? (e.g., "Professional", "Witty", "Empathetic")

Let's look at how to apply this to the 3 D's: Discussion, Decision, and Idea Generation.

1. Discuss (and Debate)

I use AI as a devil’s advocate. When I have a strong opinion on a topic—say, a new leadership strategy—I don't just want validation. I want holes poked in my argument.

Don't say: "What do you think of this idea?"

Say this (The PROMPT Method):

(Persona) Act as a critical, high-level business consultant with 20 years of experience. (Purpose) I am planning to implement a new 'Open Door Policy' for my team, but I am worried it will kill my productivity. (Request) Analyze this policy. (Tone) Be direct, skeptical, and objective. (Modifiers) Do not give me the benefits. I only want the risks. (Output) Give me a list of the top 5 potential failure points and one counter-measure for each.

Why this works: You forced the AI to stop being "nice" and start being useful. You gave it a specific lens (Consultant) and a specific constraint (Risks only).

2. Decision Making

We all get decision fatigue. Sometimes you just need to weigh options objectively.

Don't say: "Should I hire someone or outsource?"

Say this (The PROMPT Method):

(Persona) Act as an objective Operations Manager. (Purpose) I need to solve a bottleneck in my content creation process. (Request) Compare two options for me: Option A is hiring a full-time junior editor. Option B is hiring a freelance agency. (Modifiers) Consider costs, time-to-train, and reliability. (Output) Create a comparison table with columns for 'Pros', 'Cons', and 'Estimated 1st Month Cost'. Follow the table with a final recommendation based on long-term scalability. (Tone) Analytical and data-driven.

Why this works: It won't make the decision for you. But by asking for a table and specific criteria (scalability), it lays out the data so clearly that the answer usually jumps off the screen.

3. Idea Generation (Brainstorming)

Writer's block is a myth. You just need a spark. When I’m stuck on a blog post, I don't ask the AI to "write a post." That usually results in robotic garbage. Instead, I ask for angles.

Don't say: "Give me ideas for a blog about burnout."

Say this (The PROMPT Method):

(Persona) Act as a viral content strategist and psychologist. (Purpose) I want to write a blog post for medical professionals who are tired of generic wellness advice. (Request) Brainstorm 10 unique, unconventional angles on the topic of "Burnout". (Modifiers) Strictly avoid cliches like "do yoga" or "sleep more." Focus on psychological shifts and leadership failures. (Output) A bulleted list of 10 catchy headlines, each with a one-sentence summary of the concept. (Tone) Provocative and insightful.

Why this works: You explicitly banned the bad results (cliches) and demanded a specific outcome (unconventional angles). You’ll likely get 9 bad ideas and 1 brilliant one. That’s all you need.

Why AI Feels “Smarter” to Some People

It is not intelligence. It is intentional usage.

People who get high value from artificial intelligence:

• Treat it like a collaborator

• Feed it context gradually

• Set clear expectations

• Use it repeatedly, not randomly

Over time, responses feel personalised—not magical.

The Toolbox: Which AI Should You Use?

They aren't all the same. Here is a quick, non-technical breakdown of the big three players right now:

• ChatGPT (OpenAI): The creative generalist. It’s great for conversational brainstorming and the PROMPT framework works exceptionally well here due to its strong reasoning capabilities.

• Claude (Anthropic): The writer. If you want nuances, better tone, and less "robotic" sounding text, Claude is often superior. It follows complex instructions (like the PROMPT framework) very strictly.

• Gemini (Google): The researcher. Because it’s plugged into the Google ecosystem, it’s fantastic for decision-making prompts that require up-to-date information.

The Next Step

Here is your homework.

Copy one of the "PROMPT" templates above. Open your AI tool of choice.

Fill in the brackets with a real problem you are facing today—not a hypothetical one. Whether it's a difficult email you need to send or a strategy you need to critique.

See if the answer improves.

artificial intelligence is only as good as the person wielding it. Stop using it like a hammer, and start using it like a scalpel.