MORE SLOP ON INTERNET (DOOMED)

Internet is doomed, there's more and more AI slop uploaded everyday as a SAAS trying to show how it is solving some kind of problem but in reality it is the clone of one another.

We are in this phase where we are seeing a sane level of insanity where we can see people writing code to solve some kind of problem which doesn’t even exist in real life.

The amount of AI we’re seeing over internet in 2025 was not in my bucket list, github is full of AI slop people are pushing AI slop over github and microsoft is training their model over the same code, there’s some kind of metaheuristic multiverse reinforcement learning happening over the internet which can predict very easily.

The internet is already doomed whatsoever you cope with people the cs majors will be the first ones to get laid off. And we can see the trend, we are in the month of august and media estimate is approx 140 (thousands), are already layed off in the Year "2025".

I remember the good old days when Github Copilot was actually good. It was the year 2023 when Vscode had the copilot integration and i got introduced to the Github’s chat box in Vscode, I vibecoded by whole vim config using the Copilot, It was not the good experience but I remember it being soo bad that it couldn’t even declare the lazy configs by itself. But we are at the phase now where people can just open the “IDE’s” and code the whole product, and yes this is true there are people vibecoding for freelance. They are pushing slop as SAAS “SLOP AS A SERVICE”.

All websites are looking the exact same. And it is not due to some kind of UI library revolution like shadcn or nextjs theming, or the loss of creativity. It is only due to the data these LLMs are being trained on “the more you push the identical dataset the more it generates the identical output”. The more you push the “framework specific” “library specific” dataset or the ideas, the more it generates the same thing again and again.

That's the reason why LLMs are bad at system design or the more niche work, you cannot get some AI model code zig code for its latest 0.14.0 branch, it will surely generate the random code assuming zig==rust. This is the place where the LLMs are currently lacking behind the more code available on the internet the better it generates the output for the specific language. But I believe this is not what we should care about, the pace at which Artificial Intelligence is growing, it is getting much better labelled dataset day by day by user input either by their chat service or the github opensource codebase.

IT IS PROMISING< THE FUTURE OF COMPUTER SCIENCE GRADUATE IS DOOMED START LEARNING TO COOK, OR YOU"LL BE UNEMPLOYED. OR ATLEAST BUY SOME LAND IN VILLAGE ITS TIME TO GO BACK TO FARMING.