Marthie's Journal

AI is eating engineers for breakfast

The world is turning on engineers.

By the world, read, ‘the tech industry’.

If you haven’t been chronically online, here’s what is really going on.

AI happened, and it’s one of the most exciting shifts in technology and software development. But it has a ripple effect; suddenly, AI can code and replace engineers.

But before AI, we spent decades putting engineers on a pedestal, paying them more than the average white-collar Joe, idolizing them to a god-like status. Now, companies are realizing that they don’t need a team of 50 engineers, and AI coding agents can do the work of 20 engineers in minutes.

There have also been some hallucinations with engineering job titles over the years.

First, it was 10x engineers.

Then, cracked engineers.

Product engineers followed.

And now? Forward-deployed engineers.

I have been watching and reading most of the conversation online, and I think companies just want software engineers who think like product managers. They no longer want an engineer who just writes code, but someone who decides what should be built, how it should be built, and what happens after it is built. Someone who thinks about the customer first, manages technical debt better, and actually contributes to the company’s bottom line. And my favourite, someone who has taste.

Which makes me stop and think, who were companies hiring before?

The way I see it, these companies built the culture of ship fast, break things (write the code now, worry about it later), and then overhired for engineering roles. Now, they are trying to self-correct and decided to make hired engineers take the blame. That’s just my theory.

But let’s think about this critically:

Can AI actually replace engineers?

Yesterday, Block laid off 4,000 of its 10,000 employees, stating that AI and smaller teams could do their work. As usual, there were the sounding alarms of AI layoffs, engineering winter, and the like.

Maybe this is just the same alarmist, doomsday talk that happened when AI first got mainstream and ChatGPT was launched. Everyone went on about how AI would replace writers and customer support teams, only to turn around and hire real people back, a couple of months later.

We said the same thing about movies and actors being out of jobs.

I remember when Google Gemini launched text-to-image and video generation. We started saying that photographers, filmmakers, and video editors would be obsolete. Today, the noise has died down. Nothing very significant has been done with this, except product demos, AI slop memes, or those jarring YouTube ads.

People who think that AI would replace engineers are, plot twist, not engineers. They have never written a line of code. They probably used Claude Code to generate a webpage, and it all seems revolutionary to them. I mean, I understand it. I’d be gobsmacked too.

We’re missing context

Anthropic’s CEO said in an interview that software engineering will be automatable in 6 to 12 months.

Yes, with AI, work that used to take hours can get done in minutes with a simple prompt. Sure, you save time and ship quickly. But how did that fast work make you feel? Nothing right? It (AI slop) rarely invokes a feeling; it just looks like anything else on the internet.

There’s also a knowledge gap happening. Programmers used to find joy in debugging and writing lines of code for hours to solve a specific problem. Figuring out what happened and building a knowledge bank of how to solve similar issues in the future.

Now, (younger) engineers can spin up a coding assistant to build an app, but the collaboration often feels detached. They don’t truly understand what happened under the hood; how a bug was solved, the considerations for edge cases, or how improvements were made. The job is done faster, but they haven’t learned anything through the process.

The more people rely entirely on AI, the more creative problem-solving and deep expertise decline.

If you don’t know the fundamentals, you’re unable to tell the difference between average and quality software. Instead, AI will amplify your ignorance. You’ll ship faster, but it won’t make you a better engineer.

The idea that we can replace enterprise-grade systems with code generated by an AI coding agent feels a little too dystopian.

Context will matter, always. And AI can’t replace human context. Imagine this: an engineer with 24 years of experience in tech, that’s 2 decades, is laid off. And the industry context and experience are gone with him or her. Who is even sharing that needed context with the 19-year-olds we’re replacing them with?

Your AI agents can only do so much, no matter how much context you feed them. Your vibe-coded apps might take you a couple of hours to build, but they can’t handle infrastructure strain as a software built with production-level quality would.

Think of when Virtual Reality (VR) launched. Meta was so convinced that we’d get to a place where we would exist purely in virtual worlds and physical, human interactions would be obsolete. It all seemed revolutionary. The media called it the next best thing. The same thing is happening with AI, too; there’s the fear-mongering and articles preaching the doom of engineers with AI. Remember, while AI is great, these companies are also trying to increase usage and turn a profit.

Here’s the meat of what I’m saying:

Take a second to think about this.

I’m not anti-AI. I use AI tools and have built AI-powered workflows to automate my tasks. However, beneath all the noise, I think engineers should use AI as an enabler, not a disruptor for their work.

AI can write code, but it can’t replace the humans who know why that code matters, what it breaks, and what should exist in the first place.

And we should treat this new engineering shift with a little more kindness and maybe relax.

Disclaimer

This is a mind dump; I’m not an expert in software development, nor am I an engineer, so I’m open to having a nuanced conversation and learning new perspectives.