It's been almost five years since I've posted on this site. I thought I'd start again by cross-posting my newer articles from my Twitter
Original article link: https://x.com/IroncladDev/status/1997120236777013271
Hopefully you'll get something out of this one. Enjoy.
Three years I've waited for AI to show its potential and practicality. Today, I watch in horror as it consumes and lays waste to the industry I know and love.
With great compute comes great responsibility, and the people with the most compute are anything but responsible.
An infection, not a bubble
Some people refer to the AI hype as the "AI Bubble" due to its rapid rate of expansion. I instead prefer to describe it as an infection.
If you paid attention during biology class, you will remember that parasites, viruses, and bacteria need a host to feed on before reproducing and spreading. The term "Bubble" seems inaccurate, because most bubbles are inflated with a single, external source.
Pictured: the AI bubble conspiracy theory illustrated by a flat earther
Despite being a major cornerstone of the industry, NVIDIA itself isn't largely responsible for aggresively shoving AI into both consumer and developer software.
The Toll on the Software Industry
You all have seen enough people glazing and sugar-coating AI, so I will not be taking a moment to do so.
Note that I use the term "Software Industry" instead of "Tech Industry". This is because I do not build rockets, solder electronic components, or work as a slave in a bauxite mine.
The Job Market & Workplace
Disclaimer: the following is my observation of people who work in software-related companies that advocate for and heavily use AI. This is not financial advice
The most obvious and undeniable side effect of AI being made publicly available is the declining job market and workplace.
Recruiters have started to leverage AI for filtering out candidates with boring resume pages. To combat this, job applicants started using AI to build their resumes. This back-and-forth quickly devolves into a tiresome battle of prompting on both ends. As a result, the bar for entry level software-related roles rises significantly higher.
Lazy Employees who try to divert their workload to LLMs eventually end up de-valuing and ousting themselves as a leech on a salary. In the eyes of the average corporate overseer, the only difference between a local prompter and a foreign bangladeshi prompter is that one costs 1/4th as much.
When a Non-technical Manager is paired with vibe coding tools, the result is usually a violent chemical reaction (dopamine overdose) that ends up with heavier workloads, people getting fired, and higher expectations.
In the end, managers are essentially end users with elevated context and permissions.
Marketers have become addicted to using the term "AI". Failing to mention the term gives them the impression that their social media post or campaign won't appeal to the masses. It seems that most of them put more trust in a two-letter abbreviation they don't understand than their own marketing skills.
You can also thank marketers for the increased number of advertisements crammed into every possible area in consumer software.
The Decline in Software Quality
The average quality of software is declining. The tools you once knew and loved reward you with ads and AI features you never asked for.
"Software Quality" is subjective. Multiple factors such as implementation, code cleanliness, and ease of use come into play. There are also different kinds of software which I will try to cover here.
Mainstream Consumer Software is not by definition "good software" and will never be. The average consumer cares only that the software gets the job done. An intrusive AI banner is just another thing to complain about.
There's no point in having hope for mainstream consumer software. No matter how advanced AI gets, I say with confidence that there will never be a version of Windows that is considerably fast and doesn't crash all the time.
Buckle up, it's only going to get worse.
Enterprise Software is a dumpster fire that is usually beyond saving. By the time a company starts offering services to businesses and organizations, it means that they weren't profitable from individual customers alone.
After switching to a more profitable customer base (primarily consisting of micromanagers), enterprise software companies begin to taylor to them by adding a bunch of unnecessary low-impact features.
One thing enterprise companies are very good at is adopting AI, which is not a good thing. People who work in enterprise companies are probably being bombarded by AI workflows, AI-generated demos, AI internal tools, and forced to work on AI features targeted to the customer base.
Once an enterprise company pivots and makes AI a cornerstone of their service(s), it usually means they are about to do a round of layoffs and slowly die out.
An example of this is JetBrains, a company that started out with a mission to build a good IDE. They pivoted to enterprise and ultimately ended up breeding AI Agents with the descendents of IntelliJ Idea. The only legacy they left behind is JetBrains Mono; the best font for programming in Neovim.
The Software Market is completely saturated with vibe-coded slop. You can't scroll Hacker News, Product Hunt, or pretty much anything for that matter without running into something related to AI. Search engines are clogged to the brim with AI-generated websites gaming SEO. A lot of investors won't take interest in a project if "AI" or "Agents" doesn't appear in the landing page header.
Large companies like Github and Vercel sort of fall in between consumer, developer, and enterprise software. They are very centralized, and try to use vendor lock-in techniques to keep you hooked.
After the word "copilot" was trademarked by Microsoft, Github seems to have taken up the side gig of shoving it down peoples' throats via ad banners and inserting it into commonly-used UI elements. The team behind the Zig Programming Language got fed up with all this tomfoolery and migrated to Codeberg for good
The Vercel platform itself isn't as aggressive when it comes to ad banners, probably because they realize the obvious fact that it turns people off. However, they are definitely leveraging their monopoly on Next.js and other open source projects towards vibe coding.
A Parasitic Cancer
Everyone hears about how AI can be used for empowering humanity, automation, and replacing difficult jobs. All I've seen is talk, hype, and big corporations taking interest without any promises kept.
The entire existence of the AI movement revolves around mass adoption and money. The movement is exploiting the Software Industry for every last penny, and it will stop at nothing.
The people pushing for AI do not strive to make better software.
They do not care about the software engineers that they depend on.
They even have the audacity to compare programmers to an LLM with write access and a shell, and are actively trying to replace them.
Vibe Coding
Vibe Coding has spread across the entirety of tech like a wildfire and continues to assist humans spit out the most horrific code known to man.
Chances are, if you're using an AI agent to scaffold something, you probably don't have the intention of building good software in mind.
I don't really concern myself with programmers who do vibe code. I just won't review their code.
I use the term Vibe Coders to describe non-technical people who refuse to learn how to code and use vibe coding tools to derive dopamine and clout. People who refuse to put in the blood, sweat, and tears it takes to do something really shows how dedicated they aren't.
If you task someone who doesn't code with vibe coding their dream SaaS startup, you will only ever see them request new features, request UI changes, and fix bugs. The end result will be a bloated application that does a lot of different things poorly.
Lastly, I am a strong believer that new programmers should not use vibe coding tools. I would encourage new programmers to read the manual, write down the examples, run them, and experiment with them. I also encourage using an AI chat to ask questions about how things work along the way if it's unclear.
What can you do?
I want to conclude this article with a call to action instead of just dumping my thoughts out and calling it a day. Just by changing your mindset a bit, you can make a difference.
- Make it a goal to write good software that lasts for years to come
- Find a skill you're good at, refine it, and shoot to become the best. The backbone of software is upheld by people who do one thing well
- Share good software with others
- God gave you a unique mind that nobody can replicate. Use it.
Unfortunately, the AI slop will continue
But well-written and well-maintained software will prevail
And you. So will you.
Top comments (1)
Great article. Thanks for sharing.
Great advice.
Keep on posting.