Recently, vibe coding (i.e., having AI write your code) has become pretty popular simply because it's convenient and fast, but there are major issues with it that make it bad for both the vibe coder and the developer community as a whole.
These are my top reasons:
App markets (e.g., App Store, Play Store, Flathub, Microsoft Store) and FOSS pull requests are being flooded by AI slop that is likely insecure, buggy, and inefficient.
People with no idea how computers work will likely "make" code that is nonsensical and buggy by being unable to sanity-check the AI or by directly asking for code to function in a way that is insecure or even abuses outside servers (e.g., scraping).
Likely impossible to copyright or license the code in any way, just like AI generated images.
Will cause smaller code hosting sites (e.g., SourceHut, Codeberg, some Gitea instances) to implement extreme bot protection to stop their servers from being overrun by thousands of companies scraping them to make their own models. And even then, it will still probably cost them a significant amount of bandwidth.
The "creator" of the code may or may not get sued if the code the AI model was trained on was licensed under a copyleft (e.g., GPL, AGPL) or non-commercial (e.g., SSPL, CC-BY-SA-NC) license, or even leaked proprietary source code, especially if it directly copies snippets of that code into the creator's code.
There is no reason why anyone would hire a developer that only knows how to code with AI, because just skipping the middleman and buying a subscription to one of these LLM services will cost at most $6,000-$7,000 per year instead of the hundreds of thousands that a developer can cost. The only reason why an employer would hire a developer is because they can do something that the AI cannot.
Developers should actually learn how computers and code work using traditional methods (tutorials, Reading The "Fine" Manual, etc.) instead of fully relying on AI.
Developers should use AI to speed up workflows with features like autofill and troubleshooting issues, rather than having the AI fully build the application with little to no developer interaction with the actual code.
I would recommend just using the tried and tested method of starting with an easy language like Python and reading and watching hands-on tutorials, slowly building the project that you have in mind, and looking up and reading the documentation when you do not know something. Continue until you have made your project, then repeat this with whatever other project you have in mind. Soon you will be a great Python dev. Then repeat these steps with whatever other language you wish to learn; you will find it much easier than last time, as you will not need to learn as many new concepts and instead mostly have to learn the new syntax and libraries.