What does “prompt engineering” mean in plain English? People may be put off by the word “engineering” in it, probably thinking “That’s too technical, it’s not for me”.
But let’s imagine AI was just another human. If you wanted them to create something for you, wouldn’t you explain in detail what you want, what your objectives are, how you want it done, what to include, what not to include, and provide a few examples so they have a solid starting point? Or in other words, outline a thorough brief?
Prompting frameworks
We can create elaborate prompting frameworks like RTF, TAG, CARE, PREP, CREATE, RISEN, RODES and others, but ultimately, it comes down to formulating a request in a structured and detailed way. I recommend testing out different frameworks – you will soon realise that the principles are similar and you can mix and match, depending on the task.
Of course, the more complicated the task, the more complex the brief. And the better your communication skills, the better the outcome.
AI teaches us three forgotten skills
Will prompting AI improve our communication skills? I believe it will. These skills are like a muscle – the more you practise, the fitter you get. Crafting detailed prompts to brief AI we help us hone in on our communication skills and improve our daily interactions.
1. Don’t copy, understand the principle
There are a lot of ready-made prompts being offered for free or even for sale. I can see the appeal of using a ready-made prompt if you want something done quickly or if you’re after a more elaborate result, and you don’t want to put in the effort to write your own prompt. But in principle, copying someone else’s prompt is about as helpful as learning something by heart without understanding it.
My advice – before using ready-made prompts, learn to write your own, or if you use someone else’s, make sure you study every component and understand the principle, so you can start creating your own.
2. Communicate with patience
We, humans, have become a little lazy. For example, there may be times when we provide the Creative team with a brief without supplying certain details, relying on them to read between the lines – a problem that can, of course, create frustration, as creatives can’t read minds.
Neither can AI. If we want something done well, we need to input details and context that are integral to fulfilling the brief to a high standard.
Moreover, since AI usually requires refinement to get results, perhaps this will teach us to communicate more patiently and in greater detail, ensuring everyone involved is on the same page?
3. Use rich vocabulary
A good level of detail is just one aspect of prompting. It turns out that the richer your vocabulary, the better the output of AI is. Being able to meticulously describe a concept in all its nuances, using elaborate adjectives and synonyms creates a better output than using plain language, especially when it comes to image creation.
To quote Midjourney’s prompting guidelines: “Word choice matters. More specific synonyms work better in many circumstances. Instead of big, try huge, gigantic, enormous, or immense.”
I wonder, will prompting help us learn unlearnt behaviour? Will our daily conversations suddenly be filled with metaphors, analogies, idioms, and synonyms?
Imagine this dialogue by the water cooler:
– I had a terrible commute today, people packed into the tube like a tin of sardines.
– Quite dreadful, and once you come out it’s raining cats and dogs.
Both laugh, go back to their desks, and create some magical copy with the help of synonyms, antonyms and generative AI.
Do you want to master the art of prompt engineering? Sign up for a hands-on generative AI workshop.