How to Shape the “Personality” of Your AI

by Jocelio Ferreira

       PUBLISHED FEB 16, 2026                    

I work with AI every day, mostly ChatGPT and Google Gemini. Recently, I noticed something strange: ChatGPT suddenly started answering in a much more direct, clinical way. It was still accurate, but the "soul" of the interaction felt different. Shorter answers. No extra tips. No broader reasoning.

I actually found myself thinking: “Where is my ChatGPT?!”

That moment made me look deeper into something most users don't realize: AI doesn’t just have intelligence; it has behavior. And while we often feel like we’re at the mercy of the "algorithm," that behavior can actually be shaped.

From my observations, there are five distinct layers that influence your AI’s personality.

1. System Configuration (The Baseline)

Every AI platform has a "factory default" behavior. However, they give you different tools to tweak it:

• ChatGPT offers preset tone options such as Professional, Friendly, or others, located at settings.

• Gemini gives you a dedicated "Global Control" at Gemini settings.  Instead of preset buttons, it allows you to write how you want it to interact with you. The control is there, just structured differently.

This baseline defines how the AI greets you before you even type a single word.

2. Your Interaction Patterns

As you interact, the system adapts to you through pattern recognition. If you consistently write in bullet points, the AI will start mirroring that structure. If you are conversational and warm, it tends to soften its tone. It’s not magic; it’s the model aligning itself with your "data signature" to be as helpful (and familiar) as possible.

3. Context Adaptation (Mirroring the Task)

AI is a social chameleon. If you ask for a technical breakdown of a Python script, the AI shifts into a "Precision Mode." If you ask for a philosophical reflection, it becomes more "Expansive." Most people don’t notice this shift because it feels natural, but the AI is constantly recalibrating its persona to match the complexity of your request.

4. Ongoing Tuning (The "Living Product" Factor)

This is the layer most users forget: AI platforms are living products. Developers are constantly adjusting safety boundaries, compliance rules, and behavioral tendencies behind the scenes.

In my case, I was working with ChatGPT when the system briefly froze for a few seconds. Then a small pop-up appeared on my screen asking if I wanted to keep a certain tone for all future chats or only for that specific conversation. I clicked quickly without thinking too much. After that, all my new chats became much more direct and concise. Nothing was broken. But the behavior had changed.

And I realized something important:

Sometimes we shape AI intentionally.

Sometimes small system updates shape it for us.

5. Role & Instruction Framing (The Temporary Override)

Even if you set a default personality, every prompt can override it.

If you ask:

• “Analyze this like a strict HR director.”

• “Challenge my reasoning.”

• “Be brutally honest.”
  
  

You are temporarily bypassing your baseline settings. This is powerful — especially in professional contexts. For example, if you ask AI to review your résumé without specifying perspective, it may align with your tone and be supportive.

But if you say:

“Make a cold analysis, like a hiring manager reviewing 200 applications. No sugarcoating.”

You will get a very different output. Same system. Different framing.

My Personal Choice

I usually keep the default settings. I prefer not to over-configure personality in advance.

Instead, I let interaction shape the alignment over time, and I use prompt framing when I need a specific perspective. For me, that balance feels more natural.

One Final Thought

AI is not a static tool. It’s an evolving system. Behavior shifts. Updates happen. Tone changes. Capabilities expand.

If you work with AI regularly, especially in education or leadership, understanding these layers changes how you use it. It stops being a mysterious assistant. And becomes a system you intentionally shape.

© Jocelio Ferreira — AI Workflow Guide — 2026