Why ChatGPT Won’t Listen To You

I’ve been there. Whenever you try to use AI, it never gives you the answer you want.

Then, when you try to correct it, it goes even more wrong until you eventually just give up.

In this article, I’m going to let you in on the 3 steps that I follow to ensure ChatGPT always does exactly what I tell it to do.

The biggest turnoff when it comes to AI

When it comes to using AI, we start in the same position.

You hop straight onto ChatGPT, give it something creative to do, and it blasts a whole load of rubbish.

But, that’s only because AI is trained to never be wrong.?

That means it will try to give you as much information as possible and invent its own context to ensure it gives what it thinks is a ‘perfect answer’

The way you need to think about AI is as an idiot with an amazing memory.

How stupid is AI?

Imagine you’re speaking to somebody with no common sense but all the knowledge in the world.

They can recite any date or fact instantaneously, but the second you ask them to put those facts into a sentence or have an opinion on the information, they completely crumble.

This is because AI doesn’t know what to do with all the information that it has, despite the expectations people have of it.

And, that’s why I call this the expectation collapse of AI

People expect AI to inherently know how you want it to respond.

When in reality, it doesn’t.

So, as more and more people start to use AI, those who understand this fact will continue to get what they want from ai.

Those who don’t realize this, will simply try AI, dismiss it, and fall behind into the ever-growing skill gap.

If you want a chatbot to respond how you’d like it to respond, tell it how to create that response.

Let’s take a look at two examples.

Which user will receive the best response from AI?

And, which user is then going to continue to use AI and reap its benefits?

It’s the latter. However, these prompts aren’t always quick to write, and they can easily become more complicated than what I’ve written above.?

You need to approach it from a certain angle to ensure you cover all your bases and give this ‘knowledgeable idiot’ the best possible chance at outputting the correct answer.

The 3 steps to crafting the perfect prompt

Crafting the perfect ChatGPT prompt comes down to these 3 tenants.

1. Context

2. Action

3. Style

Context

If ChatGPT doesn’t know what position you’re in, it won’t know how to act. You can’t throw AI in the deep end and expect it to come out an Olympic-level swimmer right away.

You need to let it know a few things:

• What situation you’re in
• Why you’re asking it for help
• What you’ve tried before
• Why they haven’t worked and;
• What the dream outcome looks like contextually.

Only then are you giving it a fair shot at producing an answer you’re happy with.

Action

Now, you need to tell ChatGPT exactly what you want it to do, but also explain what you don’t want it to do.

This is where you outline the problem you want to solve and what the AI should do about it.

You should be as specific as possible in this step, and try to make sure that you leave as little room for error as possible for the AI. Cover some examples of perfect outcomes, but also some non-examples of what you don’t want it to produce.

Style

You need to tell ChatGPT how you want your answers to be outputted. It can’t read minds after all.

Be specific in how you want your answer to be framed and avoid any vagueness.

Do you want a bullet-pointed list? Numbered List? Paragraphs? A formal tone?

Let it know all that, and it will output it exactly as you want it to.

Okay… But 3 steps seem too much to do every single time I want an AI prompt

That’s exactly why you should look at building out systems for whenever you use AI.

You don’t want to spend so long writing your prompt that you might as well have done it without AI at all.

Connect with me or fill out this form to learn more: https://forms.gle/FQ7Uj24cQyQ2gEn3A