What’s the big deal about Prompt Engineering?

When I first heard the phrase ‘prompt economy’, I thought it meant the speed at which goods and services were produced to drive the economy. Ignoramus! Indeed, the ‘prompt economy’ is already upon us. McKinsey estimates in ‘The economic potential of generative AI: The next productivity frontier’ that the application of generative AI to different industries could add up to US$4.4 trillion to GDP annually – equivalent to adding an economy about the size of Britain!

Without a doubt, the macro transformative power of generative AI has captured the imagination of the world. But from a micro perspective, I was curious to find out how an individual like me would be able to tap generative AI. This led me on a Prompt Engineering learning journey to understand how to craft effective inputs and finetune parameters to get desired results from AI models.

And if you’re thinking this blog post was written by generative AI, think again! I wanted to put my own writing skills to use after attending the course ‘Ninja Writing: The Four Levels of Writing Mastery’. But I digress…

?Here’s some of take-aways from the Prompt Engineering course I took :

?Know your terminology:

• Temperature: this controls the creativity levels of generative AI.
• Role prompting: this is where you assign a specific role or identity to the AI to help guide the response.
• Chain of thought prompting: in this type of prompts, complex tasks are broken down into smaller ones to help solve and increase the accuracy of the desired result.

?The 3 Cs to make good prompts:

• Clarity: be clear and concise in your prompt. Start with a good action verb and be specific about the desired outcome.
• Context: provide as much detail as possible. Progress from simple details to more complex ones over time.
• Creativity: depending on the problem to be solved, encourage the generative AI to be as creative as possible [define the Temperature - see above].

?Assume nothing: It would be foolhardy to think that the output from generative AI is accurate because the answers may be wrong and/or contain biases or misinformation as AI generates answers based on what is available from its database. Though generative AI’s output is available in a matter of seconds, it’s imperative to question and validate the output each time.

?As with most skills, practice and patience make prompt engineering skills perfect because the whole process of using generative AI is iterative. You will need to experiment with different approaches and reframe/refine the prompt as appropriate to achieve the required output. ?

?In sum, I learnt that prompt engineering is both an art and a science. It combines accurate language skills, problem solving and logical thinking so that you can understand the problem, break it down and articulate it effectively to guide the generative AI to a desired outcome. Through my learning journey, I found that the collaboration between humans and generative AI makes for a powerful combination because the latter amplifies work effort by improving productivity and efficiency. However, to get the most out of generative AI, we need to be cognizant of defining the task at hand with clear prompts, critically assessing the output and further refining the prompts.

?The old adage ‘garbage in, garbage out’ aptly captures why I think prompt engineering is an essential skill for anyone who wants to thrive in the prompt economy.?