- 8 Simple Mistakes That Make 90% of ALL AI Prompts Suck 👎
8 Simple Mistakes That Make 90% of ALL AI Prompts Suck 👎
Learn the secrets so your prompts don't suck!
More than 90% of all ChatGPT and AI prompts suck.
But here’s the crazy thing…
Most prompts suck because of simple mistakes.
The mistakes I’m going to show are so easy to fix that it can immediately improve your prompts!
Here is the list of mistakes:
Mistake #1: Ambiguous Terminology
Ambiguous terminology is a massive mistake.
But, it's also one of the simplest mistakes to dodge.
In any language, there are unambiguous words (usually called objective terms) with agreed-upon meanings and other more ambiguous words where the definition varies between people.
Often, these more ambiguous terms are called subjective terms.
Let me give you an example.
Non-ambiguous (objective) words: You're wearing shoes.
If I ask 50 people, everyone will agree you're wearing shoes.
Ambiguous (subjective) words: Your shoes look good.
If I ask those same 50 people, the results will differ.
Some will say your shoes look good, while others won't.
Because 'good' is a subjective term. What looks good to one person might not look good to someone else.
Don't believe me? Watch what happens when a husband tells his wife she “looks good.” Oh boy. That's a hornet's nest!
The AI works the same way as you and I when it comes to subjective terms.
“Write me a good blog post about how to wash a dog.”
What will the AI do differently now that you've added the word 'good,' modifying the type of blog post you want?
It's impossible to know, and that's an issue because repeatability takes a hit.
When creating a prompt, your main goal is to make sure the AI produces repeatability consistent quality output.
But when you start loading up your prompt with all sorts of subjective terms, it begins to make the AI perform inconsistently.
TLDR; Avoid using too many subjective terms in your prompts.
Mistake #2: Cause → Effect Ambiguity
The second mistake is called cause and effect ambiguity, which has some overlap with the first mistake (above) of ambiguous words.
So, what's cause and effect ambiguity all about?
It's when you can't figure out which words in your prompt lead to specific outcomes. This issue usually pops up in two situations:
1. Ambiguous words
2. Fluffy sections
We've already covered ambiguous words in the first mistake. When you use vague terms, you can't tell how the AI will interpret them or what it's doing. So, repeatability takes a huge hit.
But what's the deal with "fluffy sections"?
A fluffy section is a bunch of words that make it hard to tell what they're adding to a prompt. All they do is make a prompt confusing for the AI and take away from the other words in your prompt.
Here is an example AI prompt:
I want you to act as a fancy blog post section body generator, providing me with unique and interesting content for a blog post section. I will provide you with a title and section, and you should use it to generate captivating, grammatically correct, and easy to read content that is appropriate for the section. Keep in mind that a section will be used for a blog and should be engaging and help readers understand the content of the blog post. Write a section for a blog post about a blog post titled 'How to Rank Higher on Google' and section is 'Content Creation for Backlink Acquisition' that is around 400-600 words long.
I bet you can spot some ambiguous words to avoid, like "fancy blog post section body generator", "interesting content", and "easy to read content," among others.
But it's a bit harder to spot fluffy sections.
The first sentence is fluffy: "I want you to act as a fancy blog post section body generator, providing me with unique and interesting content for a blog post section."
We don't need to tell the AI to act like a blog post section body generator.
Because we already mentioned it later in the prompt when we say, "Write a section for a blog post..."
You might not know this, but the more crap you include in your prompt, the less weight is given to the important stuff. By avoiding fluffy and redundant language, AI can deliver better output!
TLDR; Avoid using too many subjective and fluffy sections in your prompts.
Mistake #3: Chat Bias & Forget Everything
First, let me bust a myth: ChatGPT (or any AI, really) doesn't have a memory.
So, why does it seem to remember your previous chats?
Well, let me show you what's going on behind the scenes, because most of the "magic" is just an illusion.
When you ask ChatGPT something, here’s what is fed into the AI:
Your past inputs (up to a certain limit)
The AI's past outputs (up to a certain limit)
Your new question or request
Basically, it's all fed back into the AI as one big prompt. That's all there is to it.
By the way, this is also why the "Forget everything we've discussed so far. Let's start a new chat." prompt doesn't work too well! (Hint: because your previous discussion is still fed back in the next request)!
Now, before you say "who cares," remember our goal is to learn how to create awesome prompts.
I see so many people making the mistake of developing their prompts using this conversational method with ChatGPT. But the problem is that the prompt becomes biased based on the previous conversation!
Developed your prompt using a conversation? Okay, start a new chat, enter the prompt, and you'll get slightly different quality.
Because in a new chat, you're missing your past inputs and the AI's past outputs. All you've got left is the new question or request you’re giving the AI.
So, the output will be different and the quality will suffer.
TLDR; Don’t rely on your “conversation” to develop a new prompt in ChatGPT.
The Other Five Dirty Mistakes…
Want to learn about the other five mistakes?
Would you also like a 100% FREE in-depth training video that shares the remaining mistakes and also goes much deeper into all eight of these mistakes?
Could I put a cherry on top?
How about an AI newsletter that shares other prompting secrets for free?
Enter your email below to immediately unlock everything for FREE so you can dramatically improve the quality of your prompts.
Enter Your Email to Immediately Access the Video