What’s the next hot job we should all be preparing for? According to a lot of people, it’s being a prompt engineer — the wordsmithing wizard who gets better output from AI models by knowing how to ask the right questions in the right way.

A quick search on indeed.com for “prompt engineer” brings back a few dozen results today, with salaries ranging from $70,000 to $400,000 per year.  Not bad!  But is it all hype?  And is prompt engineering a real job/skill these days?  Let’s find out.

Or maybe we should start with: what is prompt engineering?  In short, it’s the art and science of asking better questions (to an AI model) to get better responses. 

For example, we could say…

…and we’d probably get a solid poem about generative AI.

But if we asked something more detailed and specific like this…

…then the model has more information to work with, and can therefore give us a result that’s more tailored to what we want.

In short, that’s what prompt engineering is all about: better input (i.e., the “prompt”) for better output.  And the input will vary depending on the use case.  For example, you could be prompting for a factual response, text summarization, text extraction, classification or conversation.  Or even prompting for code, reasoning or math.  There are lots of possibilities.  For more about the basics of prompt engineering, check out our Prompt Engineering 101 blog article.

A prompt engineer (sometimes also called a “dialog engineer”), then, is someone who writes and tweaks these prompts for the various AI models out there, like OpenAI’s ChatGPT or DALL-E-2, Google’s Bard or Microsoft’s Bing (and a host of others).  It’s their job to have a conversation with the models and get the desired results in the most efficient way possible.

In a word, yes.  And in the future, maybe no. Let me explain.

AI models are powerful, but their power is best harnessed by people with expertise in how they work.  Kind of like anything in life.  Hand me a guitar, and you’ll get a sketchy version of “Sweet Home Alabama” (sadly).  Hand that same guitar to Keith Richards, and you’ll get very different results!  The same is true when it comes to prompt engineering.  It’s indeed a skill to converse with artificial intelligence.

But is it a full-time job?  Well, from looking at recent job postings, that’s a “yes” too.  With the proliferation of different AI models, companies are hiring people to make sense of them all, and get the most from the technology.  These jobs could be for prompt engineers with general experience, or for folks with more domain-specific skills, such as in the fields of finance, healthcare or law.

It’s too early to know if this is a job with staying power and long-term potential.  Remember not too long ago when everyone in Silicon Valley was jumping ship at their “traditional” tech jobs to get into crypto, NFTs, Web 3.0 and the Metaverse?

We all know a lot of those jobs didn’t pan out.  And some prompt engineering jobs could end up the same way, for a few reasons:

• AI is getting smarter and smarter.  Today is the worst it will ever be (yup, let that sink in).  It’s entirely possible that AI will be able to prompt itself in the future, eliminating the need for much prompting from humans.  AutoGPT is already heading in that direction.

• Prompt engineering might become the next “Googling.”  These days, most people know how to effectively search using Google.  It’s just a thing we all know how to do.  Over time, maybe the same is true for prompt engineering, making it less likely that we need “experts” in this area.

My personal take is we'll up somewhere inbetween.  Prompt engineering will become part of everyday life, but we’ll still need experts who really understand the ins-and-outs of different models, how to adjust parameters (like in the Playground for OpenAI today), how to measure and improve performance, and so on.

If you have a job in tech, you’ll likely be expected to have some sort of intermediate-to-advanced skills in this area.  And since it’s early days, it’s best to build this expertise as quickly as possible to take advantage of the current wave.

In the past, prompt engineering jobs mostly went to folks with degrees in data science, machine learning and artificial intelligence.  It was just part of their role, along with actually building, evaluating or using the AI models.  

With the emergence of ChatGPT and the generative AI craze that’s followed, it seems that anyone can learn prompt engineering now.  But is that true?

Let’s break down some must-haves and some nice-to-haves for this job.

1. Writing skills

Strong writing skills will be a requirement.  After all, prompting is all about language.  This is one tech job where having a degree in English, literature or philosophy will get you far (kids, tell your parents!).

2. Experience with a variety of AI models

While some jobs might be focused on only one or two models/tools, the more you can put on your resumé, the better.  And understanding how they work under the covers is important too (such as Generative Adversarial Networks (GANs) vs. transformer-based models).

Some of the more popular tools include:

• For text: OpenAI’s ChatGPT, Google’s Bard and Microsoft’s Bing

• For images: DALL-E-2, Midjourney, DeepAI and Stable Diffusion

• For music: AIVA and Soundful

• For code: GitHub’s Copilot and Copilot X

• For video: Descript, Synthesia and Deepbrain AI

3. Coding skills

Does a prompt engineer need to know how to code?  From perusing current job listings, it is not a requirement (some explicitly say you DON’T need to know how to code).  However, in my view, coding skills will give you an advantage in this job and many others.  The most common languages are Python and R, and some job postings also require SQL. LangChain, a popular framework for working with large language models (LLMs), is also popping up more and more.

4. A college degree

Do you need a degree to be a prompt engineer?  Some companies will require a degree for any kind of job they hire for, and this is no exception.  The most commonly-asked-for is a computer science degree.  However, because it’s such a new field, I’m seeing more postings that simply require “relevant and transferable experience.”  If you can build the skills above, and have a strong resumé or portfolio to prove them, you’ll be well on your way to a job in prompt engineering.

At this point, hopefully you have a better idea of what prompt engineering is all about, and what a job might look like in this field.  It’s early days, and there are a lot of opportunities out there in this exciting world of AI.  

So get after it!  Check out these additional resources for more about prompt engineering:

• Getting Started on Prompt Engineering with Generative AI

• Prompt engineering 101 for developers

• How to use ChatGPT to write code

• ChatGPT Vs Bard: Which is better for coding?