What is Prompt Engineering? The #1 Job Skill in 2026
Prompt engineering is writing, specifically writing for a machine that thinks differently than a human reader, and this lesson frames it as a craft anyone can learn by practicing.
Is prompt engineering a technical science or a creative art? This opening lesson from Humanitarians AI argues it's neither in the way most people assume, it's closer to writing, and writing is a craft you get better at only by doing it repeatedly.
Prompt engineering as writing, not equations
The core analogy driving this lesson is simple: almost anybody can write, in the sense that almost anybody can type words into a computer. But there are very few genuinely good writers or poets, because writing well is a craft that takes real, sustained effort. Communicating effectively with a language model works the same way. It isn't about equations or technical mastery of how the model is built; it's about understanding your audience and shaping language deliberately to reach it, the same discipline that separates competent writing from writing that actually lands.
A different audience than human readers
The key distinction from writing poetry or books is who you're writing to. A poem or a novel is addressed to other humans. A prompt is addressed to a machine, and effective prompting means understanding, at a high level, how that machine "thinks" so you can shape your input accordingly. That's why later lessons in this course exist: understanding what a language model is and how it processes input directly shapes how you should write for it.
Why this is the number one skill right now
The claim behind this lesson is direct: prompt engineering is currently the most valuable skill across intellectual fields, because virtually every field is now using language models in some form. Whether the goal is drafting policy, analyzing environmental data, providing crisis support, or just getting useful output from a chatbot, the difference between someone who prompts well and someone who doesn't can be the difference between a task taking ten minutes or ten hours without producing the result they wanted.
Learning by doing
This isn't framed as a lecture course to be absorbed passively. It's explicitly a call to practice: lots of tips, lots of examples, and a lot of hands-on repetition across different use cases, prompting for code, for art, for video, for websites, and more. The instructor is upfront that even a video like this one, which looks polished, took only about fifteen minutes to produce once the underlying prompting skill was in place, which is itself meant as an example of what fluency in this craft makes possible.
Key takeaways
- Prompt engineering is framed as a writing craft, not a technical or mathematical discipline.
- The key shift from other writing is audience: you're writing to a machine, not a human reader, which changes what effective communication looks like.
- Understanding how a language model processes input at a high level directly improves how you should write prompts for it.
- The course argues prompt engineering is the number one job skill right now because nearly every intellectual field is now using language models.
- Like writing or poetry, prompt engineering is learned through repeated practice, not by reading about it once.
Who this is for
This is the opening lesson of the Humanitarians AI prompt engineering course, aimed at anyone looking to build practical AI skills, whether in high school, college, early or mid-career, or returning to the workforce, and it sets up the rest of the module's hands-on lessons on how language models work and how to prompt them effectively across code, art, video, and more.
Full transcript(auto-generated, with timestamps)
[0:00]Hey, Nick Bear here. Uh, welcome to the Hemanitarian's AI course and prompt engineer. Part of the humanitarian's mission is to upskill people. Whether you're in high school and are looking to learn from AI, whether you're in college, whether you just graduated, whether you're early into the job force, middle in the job force, later in the job force, retired and want to do something working with AI is prompt engineering. So why are we looking at a bunch of people writing to robots or just writing on a laptop? to make the analogy that anybody can write and most anybody can write. I mean they can type in a computer
[0:46]But there are very few good writers or good poets. It's a craft. If you're outstanding writer or poet, you spend a lot of time working on your craft. In order to communicate effectively with human beings, in order to effectively work with AI, you need to effectively communicate with a language model. And in the next few videos, we'll explain what a language model is, not in equations, but understanding of who your audience is, who you're writing to. The big difference between writing for example poetry and books and writing for language models is you have a different audience. A poem you're you're uh writing to humans as you are with books.
[1:34]When we write prompts we write to a machine. So we have to understand how that machine quote unquote thinks in order to write effectively for the machine. So this course is going to be a lot of practice. You learn by doing. And the starting of it here is just understanding why prompt engineering is effective and what the art is. It's a craft. It's very much like writing or poetry. You we'll have a lot of tips, but you'll learn by doing and you'll learn by practicing. The next few videos we're going to explain not again not in equations but at a high level what is a
[2:12]Language model and how do they work? Knowing how they work at a high level is critical because that shapes the way you give it input. It shapes your input to the model. So to summarize, uh this is the first humanitarian AI course. It's on prompt engineering which is the number one skill in the world right now. If you're any kind of intellectual field, they're all using language models. In order to talk to chatbt, Gemini, Claude, whatever you're using effectively, you need to be able to prompt just like you need to be able to write a clear letter if you're in business. Writing is critical. Writing for a language model
[2:53]Is probably the number one skill of the world right now because how you do it, it might be 10 minutes prompting or 10 hours prompting and not getting what you want. So that's it. This is just an analogy. Uh the analogy is this is sort of like a writ engineering course. We're going to give you a lot of tips. We're going to show you how to prompt for code, for art, how to make the videos and art that you're looking at right now, for websites, for you name it. We're going to prompt for basically everything and in every industry. But to start with, you have to
[3:28]Understand what prompt engineering is. and it's writing, but it's writing to a language model. It's writing to a computer, a specific type of computer called a language model, which we'll explain to you in the next four, five, six, seven videos what that is. Then once you understand what that is, we'll give you lots of tips and lots of examples. We'll just show off for example, I'll just make a video like this. This video took maybe 15 minutes to run and um uh just lots of examples. So that's it. Again, if I'm establishing right now what prompting is is writing just like writing sort of anybody can type, most
[4:09]Anybody can type. Very few people actually write well. Anybody can learn to write well, but it's a craft and you have to work at it and practice, practice, practice. So, if you want to get good at this, we'll help you, but you've got to put in the effort. You got to put in the practice. You got to put in the work. Okay, take care. This is the very first humanitarian AI course video. We're excited about this and many more, but we decided prompt engineering first because it's by far the most important. Okay, take care.
More videos
2:08Bridging the Pixel Gap in Browser Automation.
2:23How One Narrow Safety Rule Can Make an AI Less Safe Everywhere Else.
2:04Why splitting a chunk from its document makes it retrieve for the wrong question
4:20Three You Can Take Back. One You Can't.
2:21Why a 50-turn agent pays for the same screenshot 35 times unless it caches the pixels
1:53