Having started my journey in tech with a barebones stack of html, php, javascript, and css - learning a framework like Laravel did help me a lot in the pre-GPT days. New coders would’nt be able to relate to a time when complete projects were built in 2-3 weeks, without AI being able to do most of the algorigthm writing. The good old ‘Stackoverflow days’.
Enter GPT-3.
At first, it helped me understand what’s working under the hood with these frameworks. But, as time went on, I began to realize that if AI is going to write all my algorithms (and be that LEET-code nuthead, I could hire), why do I need to use a framework at all? You can't be serious about tech, if your dependence on a framework has overstayed its welcome.
What’s the point of Laravel or NextJS, when a basic app. py/php were good enough for a backend, and simply index.html files with a main.js and styles.css can be easily edited in deployment.
If AI is going to write all my code, the only way I can add value to the project, is by doubling down on my understanding or architecture, abstractions of information flow, and try and keep each file short enough to copy-paste within the context-window of the latest AI model.
The framework mafia, as Pieter Levels, calls it - entices you with the illusion of making stuff real easy. An illusion that works great. Unless, you are serious about shipping something useful fast.