[For Beginners] Objective-Driven Prompting

Introduction

When you talk to AI, don't you find yourself asking things like, "Tell me about X"?

I used to do that too. And the answer I'd get back would be, "Hmm, that's not quite what I meant," so I'd ask again, and it would still be a bit off... It felt like we were both just fumbling around in the dark.

But then, one day, it suddenly hit me.

AI works so much better when I tell it "what I want to do" rather than "what I want to know."

That is what we call an "Objective-Driven Prompt."

What is an Objective-Driven Prompt?

An objective-driven prompt is, in short, a way of telling the AI "what you want the final outcome to be."

With a normal prompt (an instruction to the AI), you would ask something like this.

This isn't necessarily bad, but the AI will just respond like, "Sure, here is a list of meal ideas," giving you something like a recipe website. Even if you think, "That's not what I wanted," the AI doesn't know that.

However, with an objective-driven prompt, it looks like this.

With this, the AI understands, "Oh, this person doesn't just want a list of meals; they want to cook something kid-friendly with the ingredients they have on hand." Then, it might ask you, "What do you have in your fridge?" or "How old are your children?"

In other words, when you state your objective, the AI becomes your "collaborator."

Why does stating the objective change things?

AI is actually much more flexible than we think. But because it's so flexible, if it doesn't know what you're looking for, it just gives you a "general answer."

People are the same, right?

If someone asks you on the street, "Where is the station?" you just tell them where the station is. But if they say, "I'm new to this town and I'm meeting a friend later, where is the station?" you might feel like being a bit more helpful, saying, "The station is that way, but there's also a cafe nearby, so you can kill some time there if you arrive early."

AI is the same; when it knows your objective, it tries to be helpful by saying, "Then this information might also be useful." Well, I don't know if it's actually trying to be helpful, but that's the result.

How do you state your objective?

Objective-driven prompting sounds difficult, but it's actually simple.

The basic rule is to start by saying, "I want to do X."

Just doing this changes how the AI behaves.

When you want to be a bit more specific, it's good to add elements like these.

You don't need to include everything; just being a little conscious of these is enough.

Let's look at a concrete example

So, let's look at an example of how it actually changes.

Before: A prompt asking for information

If you do this, the AI will give you a textbook-style explanation like "A book report is..." Well, it's not wrong, but that's not what I'm looking for.

After: Prompts that convey your objective

When you ask like this, the AI understands the situation and replies something like this.

See, it's completely different, right?

The AI stops being a "teacher" and becomes more like a "friend who thinks along with you." This is the power of objective-driven prompting.

Common Misconceptions

Here are a few misconceptions that beginners often have.

Misconception 1: I have to ask the perfect question

That's not true at all. In fact, it's better if it's a bit vague.

Asking like this is perfectly fine. The AI will ask back, "What kind of clarity are you looking for?" so you can refine it as you converse.

Misconception 2: Telling the objective takes time

It's the opposite.

By stating your objective at the beginning, you reach the answer you want faster in the end. It's easier to just say "I want to do this" at the start rather than going back and forth saying "No, that's not what I meant..."

Misconception 3: You shouldn't dump everything on the AI

I hear this often too, but stating your objective is different from dumping the work on the AI.

Stating your objective means clarifying "what you want to do," not saying "do it all for me." It's more like sharing your goal so that you can stay properly involved.

Tips for Objective-Driven Prompting

Once you get used to it, you'll be able to state your objectives naturally, but there are points to keep in mind at first.

Tip 1: Try starting with "I want to..."

Just adding "I want to..." at the beginning makes a big difference.

Just by doing this, the quality of the AI's response will change.

Tip 2: Add just one sentence of background

Adding just one sentence about 'why you want to do it' makes things even better.

When there is background information, the AI can provide suggestions tailored to your situation.

Tip 3: Think about 'What would be okay?'

If you communicate your final goal, the AI will work toward it.

When the goal is clear, the AI can act concretely, saying things like, 'Then let's use a simple structure.'

What if it doesn't go well?

Even if you state your purpose, sometimes you might get an answer that doesn't quite hit the mark.

In those cases, it's not that the AI is bad, nor is it that your way of explaining is bad; it's simply that 'there isn't enough information yet.'

It's fine to add more information, like 'That's not quite it' or 'Could you do it more like this?' The AI won't get angry, so you can correct the course as many times as you need.

Also, sometimes the purpose itself is vague. In those cases, you might realize 'Oh, this is what I actually wanted to do' while talking to the AI, so that's fine too. It's like finding your purpose while having a conversation.

Summary

Objective-driven prompting is, in short, just about 'telling the AI what you want to do,' but the experience of using AI is completely different depending on whether you do this or not.

AI isn't just a tool that answers what it's told; it's like a partner that thinks with you. But even a partner can't act if they don't know 'what you want to do.'

So, next time you talk to an AI, please be a little conscious of this.

Instead of 'Tell me about ~,' try 'I want to do ~, what should I do?'

Just by doing that, your interaction with the AI should feel pretty good. Probably. No, definitely.

Well, whatever. Let's just try it for now. If it doesn't go well, you can just change how you ask.

Have a great AI life!

Afterword

Thank you for reading this far.

I have summarized objective-driven prompting, where you tell the AI what you want to do.
I hope you find it helpful!

It would encourage me if you could 'follow' or 'like' this!
I look forward to your continued support!