Categories
My Opinion

AI is a Loom: The End and the New Beginning of Web Dev

Web dev as we know it is deprecated. We just haven’t downloaded the latest version yet. What comes next is a metamorphosis, a paradigm shift, a revolution changing how we work, how the web works, and everything we know.

In March 2023, OpenAI CTO Greg Brockman drew a rough sketch of a website on a livestream, uploaded it to GPT, and got fully functional HTML, CSS, and JavaScript in return. That feature – dubbed “multimodal GPT” – is shipping to all ChatGPT users in the coming months. Soon you’ll be able to open an app, provide some instructions, and have a fully designed and populated website with real content, an ecommerce store, and a social media marketing campaign built in minutes with the help of AI. 

I’m not saying coding is a dying craft; I’m saying the craft of actually writing code on a keyboard to be input into a coding environment for further processing will become a narrow specialty. Most people who write code today will not be writing code in the near future. Many of them don’t need to and shouldn’t have to write code today! Writing code instructing computers to do things when we have software that writes better code for us makes no sense, and this whole notion of everyone learning how to code feels more and more like telling highschool students their future employment hinges on their ability to master T9 texting.”

Me, in an email dated November 2021

Web development stands on the precipice of an AI-driven metamorphosis. In front of us, the demarcation line between two paradigms: The old world of human-written code aided by machines, and the new world of AI-generated code aided by humans. 

For the web, AI is the Jacquard loom. 

For most developers, this means transitioning from writing and debugging code to directing AI what to build and checking its work. AI represents a Jacquard loom moment for web development, transitioning our work from hand-coding the fabric of the web to using that fabric as material for building new experiences.

The implications are enormous, not just for our work but the web’s future. As AI becomes part of our practice, our role shifts from writing standards-based, accessible, progressively-enhanced code to ensuring AIs use the latest, most advanced standards to build the future. If we don’t embrace this new role, progress will stall as AI biases established standards and ignores new tools and best practices..

Here’s what I see:

Very soon the public will access AI services that create websites in minutes from prompts, sketches and assets. Wix teased this, and competitors aren’t far behind.

I’d be shocked if Canva and Figma don’t unveil full “draw it, AI builds it” services by year’s end. Soon there will be ChatGPT plugins that build websites for you from scratch. This is inevitable.

When I say this out loud, the immediate response is usually some version of “AI can’t write good code” or “AI doesn’t understand users” or “AI makes constant mistakes.” All true, and all irrelevant. This isn’t about AI writing code or autocorrecting our code. AI will instead use the well-documented and well-established frameworks, templates, build processes, and automation we’ve created to make our work easier to weave together the websites and apps we ask for.

For walled gardens like Wix, this is straightforward: their back-ends, systems, and design languages allow AI to rapidly wire sites to user specifications. And that’s just the start. We’ll soon see new semi-agentive tools supporting various stacks, so you (with the help of an AI) can select frameworks, design systems, ecommerce platforms, etc. based on project needs without writing or knowing how to write code.

Look at what the people over at Builder are doing, then add an agentive AI on top and you start getting the idea:

What People Want, What Automation Provides

Two massive waves of progress are converging:

Developers have spent a decade building automation tools, frameworks, and systems to improve dev experience. You can now spin up a Next.js site in GitHub Codespaces in minutes without writing a single line of code. Component-based frameworks provide code consistency and add LEGO-like block assembly to web development. Design systems, component libraries, style guides, and tokens enable rapid prototyping and consistent UX/UI. Linting, optimization, accessibility, testing, CI/CD are largely automated. Bridging layout and code is reality. Often, we just connect these pieces. AI serves well as an automated and intelligent loom weaving these pieces together into workable materials.

On the user side, people want friction-free, no-maintenance, always-on experiences. Faced with the choice between the DIY bazaar of the open web and the shiny mall of app-based walled gardens, they pick the moving sidewalk of least resistance. And they are willing to pay for that convenience; with money and by giving up their data and privacy to surveillance capitalism. Where publishing on the web used to mean standing up a WordPress site (or paying someone else to do it), today bloggers, creators, influencers, and small businesses opt for TikTok, Instagram, YouTube, Medium, Substack, Shopify, and Linktree. 

The web we lost is a bygone web a larger and larger portion of the public never experienced, and concepts like self-hosting seem archaic and inefficient to the masses. Counterintuitively AI may help bridge this gap and reignite the interest in carving out your own space on the web by lowering the barriers to entry down to describing what you want and watching it manifest.

What is pushed down as these waves converge and elevate the capabilities of the web-using public is the need for traditional developers. When the options are either an AI site from Wix built from a prompt in minutes or a complex and expensive custom build that takes months to complete, there’s no choice for most people and businesses. When the Jacquard machine automated weaving, hand-woven textiles transitioned from an essential commodity to a luxury art form, and the expertise of manual weaving morphed from a commodity skill into an artistic pursuit. Weavers still exist, and bespoke fabrics are still made, but the vast majority of textile products were made by machines guided by humans who spent their time designing the products instead of making the materials. That’s what comes next for the web. 

AI Creates Opportunity Space

This may sound like AI replacing humans. It’s not. Instead it’s a fundamental shift and refocusing of the role of the developer: From writing code to auditing AI-written code. From knowing how to wire together different frameworks to architecting the system that serves up the website. From fighting with CSS to fine-tuning AI-powered user experiences. 

The people currently working as coders will take a step up the ladder to focus on higher-order capabilities, using their expertise in languages and frameworks to help AIs produce the best output instead of doing the grunt work of writing the code. 

Web dev as we know it is dead. What comes next is a metamorphosis, a paradigm shift, a revolution changing everything we know.

Our new human focus as we move into this future together is to ease the persistent tensions found in the intersection between technology and humanity. AI can’t conduct UX research, design novel experiences, innovate standards and best practices. That was always and will remain our territory. As AI takes over the work of weaving the fabric of the web, we do the work of making new things with those materials while improving their quality and inventing new ones.

In the short term, we’ll become AI managers – customizing, configuring, ensuring user flows and information architectures make sense, monitoring the generated code to ensure the latest standards are in use, and counteracting the inherent bias of AI to repeat prevalent patterns even when they are outdated. We’ll shift from writing code to deciding what the code should accomplish. To do that, we must all become experts at the bleeding edge of code, and invest our time in innovating new standards, patterns, and frameworks for the AIs to use. It’s a whole different job needing a whole new version of the skills we’ve always had.

This transformation is happening now. For consumers and SMBs, it will be lightning fast. For institutions and large enterprises it will be slower, hindered by legacy systems, institutional inertia, and resistance to change. But it’s coming. 

For web workers, it is no longer enough to know the core languages and established best practices. UX, interaction design, accessibility, and innovation is our new bread and butter, built on a strong foundation of modern web standards and bleeding edge HTML, CSS, JavaScript.

The future of the web belongs to those who strategically apply AI to meet user needs. With proper guidance, AI can supercharge our work, provided we put ethics, accessibility, user experience, and innovation front and center.

We build the future with every decision we make. How we decide to work with AI decides what future we get to live in.

Cross-posted to LinkedIn and dev.to

By Morten Rand-Hendriksen

Morten Rand-Hendriksen is a Senior Staff Instructor at LinkedIn Learning (formerly lynda.com specializing in AI, bleeding edge web technologies, and the intersection between technology and humanity. He also occasionally teaches at Emily Carr University of Art and Design. He is a popular conference and workshop speaker on all things tech ethics, AI, web technologies, and open source.