What is “Prompt Engineering,” Really?
By Robert Walker Cohen
The AI-savvy have mostly moved on from prompt engineering and onto agents. For the masses, different story.
According to Google Trends, the term is catching fire. An uptick began over the summer and has only gained steam since, hitting several new peaks — including an all-time high just six weeks ago.
My hypothesis is that this is driven by the rapid, wildfire-like adoption of AI in the workplace. Across global corporate offices, AI is achieving mass adoption and ubiquity. Humans are curious, and like to improve, so they can be expected to naturally wander into prompt engineering. Ever-more curious youth, at university or high school, have likely ventured down their own prompting rabbit holes. It’s hard to imagine a white collar professional or student today not prompting an AI model at some point during the workday, whether for search, ideation, feedback, or some random MacGyver.
I’m not a fan of the term “prompt engineering,” as it makes out a relatively simple activity to be more technical and complicated than it actually is. Often, posts of prompts that gain some measure of virality or widespread use tend to be walls of text that intimidate non-technical or casual AI users. The complexity can be a point unto itself, rather than any real utility.
I’ve written complex AI prompts of my own and found the results to be valuable and useful, so I’m not opposed to the idea of writing a detailed prompt even if I do oppose unnecessary complexity and obscurantism.
In fact, I embrace detailed prompt writing. Writing a lengthy and specific prompt can be helpful for the prompter, as it forces clear thinking and articulation of the required outcome. Where I object, is referring to this as “engineering,” as if it is a technical skill that requires a university education.
Nothing is being engineered. Only clarified, to produce a specific and desired output. Anyone can do this.
A better term than prompt engineering, less intimidating and more casual friendly, may simply be — “Detailed prompting.”
Engineering requires a degree. It denotes exclusivity.
Simply being detailed does not. It’s a decision, available to all of us. As with writing, it starts with understanding exactly what it is that you want to say.
