Not the end of the world, not the end of programming, but the end of the web development industry as we know it. Let me explain.

 

First glimpse of GPT-3

Somebody posted this tweet during last week's TWIW. I just brushed it off because I thought it's a project of someone who's messing around with machine learning. It took me a while to realize that the technology underneath is built by OpenAI.

 

I realized that I was already describing GPT-3 in my talk, but I thought that it's going to arrive at some distant point in the future.

 

During the TWIW stream, we discussed a couple of things that GPT-3 can do (for now), but I still didn't take it that seriously... until I started reading a bit about it before going to bed. Did you read that correctly?

 

Researching about an AI that's gonna absolutely change web development as we know it - before going to sleep.

 

Worst. Idea. Ever.

 

Of course, I didn't fall asleep easily because THIS MOTHERFUCKER CAN TRANSLATE PLAIN ENGLISH TO SQL.

 

What?! Why is everybody not losing their frickin' minds over this?!?! I might know why.

 

The spoiled web industry

When the COVID situation escalated and everyone had to start working remotely, there was a tweet (that I couldn't find, obviously) from an engineer at Google who complained that now they would have to make coffee at home. I'm sorry, what?

 

Are we talking about the same old coffee that everyone drinks? The coffee that literally takes a few minutes to make? Or do you have a special Unicorn-Tears-Infused-Double-Googleccino-With-The-G-Logo-Drawn-With-Foam coffee? Wait, who the fuck are you, Madonna? The lead singer of Queen? The goddamn Queen herself?!

 

It's amazing that this industry full of rockstars and ninjas got so spoiled that we cannot even make coffee at home, so of course we got so comfortable in our positions, that we're not even scared of a goddamn AI.

 

But, it is inevitable. The rockstar web developer dream is starting to crack a little bit, and it will be absolutely shattered one day, of course, with the help of no-code tools and AI.

 

I hate when people say this, but...

 

"I'm soooo here for it".

 

We're code plumbers

Web developers get offended when someone takes a jab at their 2 favorite "languages": HTML and CSS. Noticed the quotes? I was trolling, but someone just stopped reading this article and started a tweet because they are offended by it and I'm sooooo here for it.

 

I could give a fuck if you call them languages or not, the funny thing here is the entitlement and pride people take in their webdev work. The most hilarious thing in the world is people putting "senior craftsman of wonderful web experiences" or "web engineer and ninja architect of structural and delightful web adventures"

 

HEY. You. Yeah, I'm talking to you, the person googling "splice vs slice" and "align-content vs justify-content" for the goddamn 3104813th time. AI is not gonna replace your job?

 

YOU'RE STRUGGLING TO COMBINE 4 API CALLS TO MAKE A PERFORMANT LIST OF 15 RECTANGLES EVEN THOUGH YOU HAVE SWITCHED 7 FRAMEWORKS IN YOUR CAREER SO FAR.

 

If you browse the entire internet you will literally find only 7 animations in total. Everything else is static, janky, and mostly ugly. You know why? Because the web platform sucks.

 

Our job is to glue together a bunch of random libraries that don't work together and slap them on top of a framework to achieve the same damn layout that we're building for the past 30 years. You're a code plumber, just like me, and just like everybody else that works on the web. Deal with it.

 

Oh, but instead of Bootstrap columns, you're using CSS Grid and instead of FileZilla, you're using a CI. So it's totally different now. Congrats, that makes you a senior ninja superstar developer.

 

sigh

 

another sigh

 

an audible sigh that made my dog close his eyes a little bit because he felt an incoming breeze

 

History is repeating itself

The problem is that most people don't want to let go of the way things are until it's too late. This is not a web developer problem, it's a human problem.

 

 

How about the factory lector who was super confident that he has a stable job and he was reading the newspaper every day until one sunny day the radio was invented and slapped that silly hat out of their head?

 

There are tons of similar automation examples throughout history, so it's so obvious that most of the webdev things will be automated. That's why it's so mindblowing how many web devs are ignorant of that fact:

 

(Actually, when you think about it, it's a goddamn miracle that we convinced a frickin' AI to write CSS instead of burning the entire web platform and starting from scratch, so that's a win for humans)

 

Moving on and letting go

Everyone is scared of the unfamiliar, so they're defending the way things are and they're praying they're going to stay that way. I mean who wouldn't want to continue making a shit ton of money by telling clients they are working on "internal fixes for optimizing performance" while tweaking their precious Vim setup most of the day?

 

I'm noticing this pattern in the webdev world over and over again:

 

• The first time I introduced Angular 1 to the company I worked for, everyone who mastered jQuery hated it and "didn't see" the benefits of a framework.
• The Angular 1 developers (including me) didn't like JSX at a first glance. "Ewww what is that php-looking-messy-non-separation-of-concerns-ugliness trying to be? Do you want some Bolognese sauce with it? HTML templates FOR LIFE, all logic belongs in a separate JS file!!!111"
• The SQL experts didn't want to use an ORM because it's "not performant enough"
• Any time I would show a backend developer something that makes backend & database work easier and more accessible, they would start a rant how that's not "real development" and if you want "performance" you will always have to write raw SQL.

laughs in Prisma

 

I'm not a fan of mindlessly jumping to libraries and frameworks just because everyone is doing it. But sometimes solutions to problems are painfully obvious. Most web devs are so scared to move on to libraries that can obviously solve their problems, let alone letting an AI do part of their work.

 

You don't believe me? Here:

 

Badly worded? Absolutely. I wish I could rewrite it differently. But no matter how I write it, most people won't even try to understand what I'm trying to say, and they'll start bashing it because I'm "offending" the technology that they're currently using. I was trying to say that if we would start webdev from scratch we would end up with solutions and techniques that are way more automated, and allow the user to do more by caring less.

 

I personally realized that technologies like Prisma, MobX, GraphQL, css-in-js, and JSX are automating most of the manual work I'd have to do while allowing me to care about real issues and solve real problems.

 

• I don't need to learn how to write "optimized and performant CSS" (whatever that means, according to that one author that's milking the shit out of the CSS book he wrote in 2002) because stuff like Tailwind and css-in-js is automating all of that.
• I'm don't need to learn how to manage and combine 1034813 different REST API calls, because GraphQL exists. Problem solved.
• I don't need to learn SQL for the same reason I'm not writing ones and zeros, or assembly code: there are already abstractions on top of it.

We invented layers on top of other layers to make our lives easier, and I see GPT-3 (or any other AI) as just another layer that will help me solve problems faster.

 

How AI will change web development

Getting a job

So funny that I have to make this disclaimer, but before you get mad about this part, I'm talking about the future. Pessimistically: 5 to 10 years, maybe sooner. Again, another reminder that I'm talking about the future.

 

Currently, the entry bar for a junior developer is super low. Take this as you will, but what I'm saying is that the bar will be waaaaaay higher in the future. Now you can learn a bit of HTML & CSS and next thing you know is you are sipping Starbucks on your way to your first internship at a big company, where you're gonna spend most of your time on tweeting CSS memes.

 

Welp, there goes that dream. And the AI won't say shit like "please don't talk to me because I still haven't had my first coffee today".

 

When all the basic things will be automated and some poor Alexa bot is gonna move things left and right by 1px, you can bet your ass that you will have to know machine learning and AI before even thinking about an internship. Managers and CEOS will be able to quickly execute on any idea without hiring a team of 15 people who need at least 3 weeks to agree on how to implement dark mode.

 

Launching a startup

Not so long ago you'd have to pay $10K for a landing page because there were only 3 people in your town that knew WordPress. Now my grandma can launch a landing page with a drag and drop tool.

 

Currently, you can watch an online course on Vue and in 3 months you can launch a "Tinder for Cats" startup that will make you millions. That's crazy.

 

However, not everyone can program, and many people are scared by the entire process. The shift is already happening with no-code tools, and in the near feature "coding" will be even more accessible and the gate will be finally lifted. No more gatekeeping. Everyone would be able to "code", and everyone's ideas will easily come to life, which will automatically reduce the value of

 

Getting paid for bullshit

Let's be real. Most developers are spending most of their day on bullshit. I know it, you know it, we all know it. You're probably reading this article in the middle of your workday, which proves my point. All the pointless things that we're wasting time on will be automated: formatting, linting, organizing code, folder structures, reordering things, testing, moving files around, pasting solutions from Stackoverflow, reinventing styling, jumping between frameworks every day, reading articles on some new library "just in case" etc. etc. etc.

 

Forget about it.

 

Humans are making webdev-related decisions based on million other factors that don't have anything to do with the problem at hand. The AI doesn't give a shit. It's not ego-driven. It doesn't have emotions. It doesn't care about clout or online presence. It doesn't care about hype, "thought-leaders", GitHub stars, or money from sold books and courses (wink, wink).

 

If will always choose the ideal solution for a problem without asking its buddies if it's trendy or not. Even if that solution is Angular 1. KIDDING.

 

Disclaimer: I swear to God, if one day GPT-3 opens a TikTok account just to post cringy webdev "humor" for clout, I'll quit web development and live in a random forest in Canada.

 

Embracing AI

Look, everything will be automated, but that's great news. Let's learn how to embrace it.

 

Stop playing with the Prettier config, because there are real problems to solve out there. Buying a $49 font is not gonna save the polar bears. Leave your precious zsh setup alone, your city could probably use 193413 different solutions to help the COVID-19 situation. It doesn't matter if the color of your git directory is your beloved Terminal app is highlighted with rainbows, go ahead and start a startup that solves an actual problem. Win-win for everyone.

 

If you waste your time on reinventing the wheel and solving non-existent problems all day, you will absolutely hate AI.

 

If you start focusing on what the people need and solve real problems, AI can easily become your best friend.

 

curtain falls down
me and GPT-3 holding hands while riding off in the sunset
me leaning towards GPT-3 for a kiss, as I absolutely ruin the end of this article in a hope to get a final laugh or a disgusted face from you

 

Anyway.

 

I'm talking a bit more about this in my talk: