Large Language Models are super intelligent, but also really dumb.
It's easy to look at something like ChatGPT, Claude or Gemini and just think that given that these are AI tools it's incredibly easy to get the answer you're looking for in seconds.
But unfortunately this just isn't the case.
The way we used to help over one-thousand learners understand the limitations of Large Language Models (LLMs) is this:
Imagine you're dealing with the smartest human in existence
They know nearly everything and can help solve nearly all of your problems
But
They have zero context on:
Who you are
What problem you're trying to solve
How you're trying to solve it
What type of solution you need
Now this can seem a bit daunting to start with. At the end of the day you've been told that tools like ChatGPT can save you hours of work and are the route towards productivity bliss.
And so, when it comes to writing prompts we teach our learners to ensure that the prompts they write contain the 5 key "ingredients" that'll ensure that they'll get the result they're looking for.
These are:
Context: Always try and provide background details. Remember these AI tools are super intelligence but incredibly dumb as they know nothing about what we're looking for and what we're trying to achieve.
Task: Always explicitly describe what you want the LLM to do!
Instructions: Explain how to do the task if it's not clear
Clarification: Clear up any confusion through filling in knowledge gaps
Refine: If you didn't get the result you wanted, refine and try again!
Like anything, writing prompts is a skill and one that needs to be practiced often. Build up your reps and get the practice in and you'll be a master with LLMs in no time at all!
As the AI tools we use become more intelligent, it’s more important than ever to write effective prompts.