Artificial Intelligence (A.I.) is revolutionizing the web development industry, but is it a threat to web developers? There’s been lots of talk online about A.I. taking peoples jobs. I recently watched this great video from Justin Jackson on why he thinks AI won’t replace web developers:
In short – I agree. A.I. isn’t going to take jobs. Much like No Code or Open Source didn’t.
For sure, there will be a certain level of developer that A.I. will take work from, much like the SquareSpaces and Wix of this world took a chunk out of the template based web design market, but SaaS apps? they are more complex in my mind.
I used to be a web developer, and transitioned careers into product design, so know enough about the space to have gone and tried most of the current tools out, and have found them amazing at certain things, but equally average at others (which is no surprise, given they are guessing based on a vast pool of training data, not all of it good).
What does a good product need that A.I. struggles with?
Human intuition in deciding what makes software good
Building a good product requires thoughtfulness, and decision making, and human intuition, something that Artificial Intelligence just doesn’t have.
For example, yes, you can ask an A.I. to make a pricing grid for your website. Will it look good? quite possibly. Will it effectively sell your product? That’s another question completely, that needs knowledge of your product, your market, and your customers current needs and wants, and not something an A.I. can give. It guesses, and lacks nuance.
Human intuition translates into designing a good product based on learned experience, and intelligence.
Consistency
A good well-designed product has rules, a design system, and ways of doing things, that A.I. can’t quite (yet) do well. If every time you ask it to do something it’s slightly different, that won’t work in a good product.
System level thinking/deep product understanding
Sure you can get from zero to one with A.I. But can you go from one to three? Can you reliably build a consistent product across a large surface?
Expertise in understanding how a large complex system functions
If you generate all your code, how do you fix bugs? how do you know what’s going on? It’s hard to understand if you didn’t write it, and potentially could take longer if one needs to customise it. My web developer colleagues all can both debug and fix things fast across the full surface area of a system as they have deep experience and product knowledge.
Managing hosting and dealing with security vulnerabilities
A lot of hosting companies (like Vercel with V0, and Replit) are making it easy to use A.I. to build products and deploy them, however that comes at a price, and a load of trade offs.
When using an A.I. agent to do things you will always be one step removed, and layers of abstraction always make it hard to do things.
What are A.I. agents good at?
Things with words and copy
At least a first draft of things. It’s always good to use A.I. for inspiration, but do the writing for yourself in your own voice.
Prototyping
They are great for prototyping – most code tools like Replit, Github Copilot and WindsurfAI (my current favourite) have moved to fully Agentic prompting (so you say “I want to build a product with X features using Y framework and Z CSS library”) and it goes off, writes a plan, and builds something step by step.
However, they build without much context of what you want, what user needs are, and how maintainable things should be going forward. As we know, artificial intelligence isn’t actual intelligence, it’s just predicting what goes next, based on a corpus of previous knowledge.
Intelligent code assistance
It’s incredible seeing what an A.I. enabled code editor can suggest, filling in whole functions, and speeding the pace of writing code. I can only see that increasing.
Going from zero to one for small personal products
I’ve built myself a small mobile-optimised editing tool for my eleventy static site. I’m typing in it now. No one else has to use it, just me.
I used Django to scaffold out the site, then used A.I. to try our different CSS frameworks, look at them, poke the code, and discard them very fast. I also used A.I. to write integration scripts for some A.I. APIs to build a chat widget in my editor.
Explaining how existing code works
What I love using A.I. for is explaining how things work – it’s like a helpful teacher that never gets tired, and can explain why or how a certain framework function works like it does, often with examples from the docs. The key (for me at least) is I understand every line of code that has been generated, and have gone in and tweaked things when they weren’t right, or could see obvious errors.
I see A.I. a bit like the mech suit in Aliens. Super powerful, but you need to know what you are doing, and if you don’t you make a hard-to-maintain mess.
Generic simple products
A.I. is good at building basic CRUD apps. However I wouldn’t trust an A.I. with thoughtful user experience, fixing edge cases, reliability, and so forth. Generally it’s great for a v1, but you often have to ask it to redo lots of things, and it lacks nuance or experience. It’s not intelligent.
So what is in danger from A.I.?
Simple generic single use products
Things that don’t need a great user experience, but just need to work, often used by a single person.
NoCode solutions might also be in danger, as these often have quite generic basic functionality.
Scripts and simple apps to perform single tasks
I’ve got friends who work in the ecommerce space, and have saved days in man hours getting A.I. to write conversion scripts to migrate products into Shopify from other platforms.
Low quality generic template websites
Basically anything formulaic that can be churned out. Probably a lot of outsourced work where things are built to a spec, and there’s not much care or iteration.
In conclusion
There’s no doubt A.I. is changing the face of the web development industry as a whole, but certainly for the short term, I think web developers (or a least the good ones) will be OK.
I think A.I. will certainly help people write code, and even debug problems at an exponential rate. As code tools improve, A.I. will augment, but not take over web developer jobs, enable more efficient working, freeing up time spent on more mundane tasks to things with more problem-solving and strategic thinking.
A.I.’s capabilities are limited in that it requires guidance and understanding from human users to deliver optimal results in product creation and design.
If web developers focus on delivering good quality work thoughtfully that provides genuine value, and learn how to use these new tools to their benefit, they will always be one step ahead.
