Be the Boss: Why You Still Need to Learn Web Development in the Age of AI

Imagine you’re a designer working at an agency. Your manager came up through sales, and doesn’t have a design background. Even though she’s great with clients, she doesn’t understand the basics of design. Whenever she wants something changed on a site, the instruction is always the same: “Just make it pop.”

That phrase tells you nothing. Pop compared to what? You guess. You bump the font size, brighten a color, add a photo. She glances at it for three seconds and says “no, that’s not it,” unable to explain what “it” was. She isn’t a good boss because she hands out instructions without understanding the job underneath them. You may have your own frustrating experience working with a boss who knows less than you.

Here is the parallel: In the age of AI, YOU are the boss. You need to give AI clear, informed direction. You’re the one in charge now, and you will be a more effective boss if you know the job.

A Good Boss Knows What to Ask For

AI has limitations. It doesn’t hand you a list of things you forgot to consider, and it won’t volunteer a fix for a problem you didn’t know you had. In working with AI you need to have understanding and take responsibility of every step AI takes. You also need to understand what it does well and what limitations it could run into.

Let me provide an example from my own work experience. I run many websites. A well designed web page should have for a solid meta descriptions. Meta descriptions usually appear in search results, and are a core part of helping potential website visitors find your website. Someone vibe coding a website without my background wouldn’t think to include meta descriptions, and their pages would quietly underperform in search results with no obvious symptom to point to.

Here are the steps I take:

1. I have AI comb through my site for pages that are missing a meta description.
2. I ask AI to write a draft for each.
3. I review the meta descriptions AI writes for me, to make sure it is compelling and accurate. Where necessary, I make changes to the draft meta descriptions.
4. I ask AI to install the meta descriptions.
5. I check to make sure they are successfully installed.

As you can see, AI saves me a lot of work by finding the missing meta descriptions, writing drafts for me, and installing them. I don’t just take whatever AI hands back. I read every draft to make sure it’s actually compelling and on target, not just technically correct, and only once it earns its place do I have it installed on the page.

That’s the whole loop: know what to ask for, check the work, then direct the fix.

Knowing what to ask for is only half of being a good boss, though. The other half is being able to tell whether what comes back is actually right, and that only comes from experience. You can’t ask an AI to fix a problem you don’t know exists, and you can’t catch a wrong answer you don’t have the experience to recognize, the same way that manager could never ask for something better than “make it pop.” Learning HTML and CSS is largely learning the list of things that matter, and building the eye to tell good work from bad.

Sometimes You Need to Fix It Yourself

Here’s the part that shows up every single day of real work: many fixes are small. A padding value, a color, a breakpoint that needs adjusting. If you know CSS, that’s a five-second edit. If you don’t, it’s a whole round trip: write a prompt, wait, check the result, notice it changed something else you didn’t ask about, write another prompt to fix that. Multiply that by every small tweak on a real project, and the “I don’t need to learn this” plan turns out to be slower than just learning it.

There’s also the moment every developer eventually has: a site goes down, or a client calls with something broken, and there’s no time to describe the problem to a chatbot and wait for a response. You open developer tools and fix it. That skill doesn’t come from prompting. It comes from understanding how the pieces fit together.

This Is Why Students Still Type Every Line by Hand

In my own classes, I require students to write HTML and CSS by hand, tag by tag, property by property. Not because of tradition, but because typing it yourself builds a different kind of knowledge than watching an AI produce it. There’s a well-documented effect in learning research: people remember material far better when they produce it themselves than when they read it or watch someone else produce it. Typing <div class="card"> fifty times leaves a different trace in your memory than watching an AI generate it once, even if you followed along perfectly both times.

The errors matter just as much as the successes. When a student forgets a closing tag or mistypes a property name and the page breaks in front of them, that’s the actual lesson, not a detour from it. It’s how you build a working, gut-level sense of the rules instead of a vague memory of having seen them once. Skip straight to AI-generated code that already works, and you skip the exact moment where that understanding gets built.

Learning anything worth having involves some discomfort: sitting with something you don’t understand yet, getting it wrong, and coming out the other side with an actual skill instead of a shortcut. That discomfort is the point, not a flaw in the process. The goal of school was never the piece of paper at the end. It’s the career after it, and a student who fakes their way through the discomfort with AI walks into that career without the skills the job actually requires. The gap doesn’t show up on the diploma. It shows up on the first real deadline.

I won’t pretend the jump from that to directing an AI is a small one. It’s honestly the biggest question I’ve had to sit with as an instructor in years: how do you keep students doing the slow, hands-on work that builds real understanding, while still getting them ready for a field where AI is genuinely part of the job? The answer I’ve landed on isn’t to choose one over the other. It’s sequencing. Foundations first, by hand, errors included, because that’s where the judgment in the rest of this post actually comes from. AI second, once a student has enough of that judgment to direct it instead of just hoping it’s right.

I’ve Watched This Happen Before

Here’s a prediction, and I want to be upfront that it’s exactly that: my own opinion, based on thirty years of watching corporate technology land in real workplaces, not something I can back up the way the rest of this article is backed up.

Early in my career, I worked at a call center right as PCs were becoming standard equipment. Management was confident staffing would shrink, since customers would just look up their own answers instead of calling in. It didn’t happen that way. Customers still called, because they had real problems that didn’t fit a script, and those problems still needed a person who understood the systems well enough to actually solve them.

I think AI is following the same pattern: not eliminating the work so much as moving it. The genuinely tedious parts, typing the same boilerplate for the hundredth time, memorizing every property name, are shrinking, and that part is real and worth celebrating. But new work is taking their place: mapping out a solution for every kind of problem a customer might actually have, then overseeing what the AI does with it, and tweaking or intervening when it gets something wrong. That’s arguably the more interesting half of the job, the half that makes a workday fly by instead of drag, which is a lot of the actual difference between a job and a career.

My honest guess, and again, it’s a guess, is that IT work doesn’t disappear over the next decade so much as it shifts toward exactly that: people who understand the fundamentals well enough to train, guide, and correct AI systems instead of doing the repetitive work by hand. In our field, that still starts with knowing HTML and CSS. It’s just aimed at a different target than it used to be.

AI Is a Tool. You’re Still the One Who Has to Know the Job.

That’s what a bad boss looks like: bad instructions going in, no ability to catch bad work coming out, and nothing to offer when it breaks. None of this is an argument against using AI. I use it constantly, and it’s made me faster at real parts of the work. But it’s a tool you direct, not a replacement for understanding the work in the first place. The people getting the most out of it aren’t the ones who avoided learning HTML and CSS. They’re the ones who learned it well enough to know what to ask for, to check whether the answer is actually right, and to fix it themselves when it isn’t.

That’s still a skill worth building on purpose, not something you can skip because a tool can type faster than you can. If you’re ready to start from the actual fundamentals of HTML and CSS, that’s exactly where our lessons begin.