Sure a Hello World app might be do-able by AI but any complex business logic is going to be done incorrectly. I work on Rails and how many times do I use scaffolding? small prototypes is about it and those don't live long.
There are lots of developers that don't like IDE's, like myself. I've been burned way too many times by them doing things I didn't intend. Unless it delivers on all its hopes and promises, AI tools will end up being a neat trick and then be a distraction.
There are some shops that use bleeding edge stuff and then there are probably like 90% of shops that stick with what works. Think of all the software developers working at banks, insurance companies, Fortune 500, etc. In 5-10 years developers will mostly still be doing the same things they are doing today.
- Something else will replace Agile. I'm afraid it will be even worse (some kind of movement driven by people who want to sell books and conference talks), but on paper it will look way leaner than current Agile practices
- By 2030, sexy third-party services that we use today and that take some core functionality of your business to the cloud, will look like legacy stuff that is very hard to maintain (more than the usual legacy code that you own). Hopefully this will lead to less SaaS and more "let's build a library for that" or "we actually don't need that, let's talk to the client"
Here are some reasons why I think this will be the case:
- Codex is an extremely low-effort first attempt but it already works for simple programs. It's capable of generating syntax errors, yet this shouldn't be technically possible if it's language-aware (it should know exactly which tokens are illegal). They literally just trained a language model on source code with no considerations for the domain. This means there is a lot of room for improvement, even without advances in ML.
- Codex is not programming as humans do, it's actually doing something much harder. Most software (for end users at least) is designed to display 2d graphics, text and interactions, yet Codex knows nothing of this entire modality. This means its capabilities should increase massively once these modalities are incorporated. It's clear that OpenAI is already pursuing this direction with CLIP/DALL-E
- Source code is actually not a natural representation of software programs, but an abstraction for humans. A more natural representation of software is the syntax tree in the compiler, or maybe a finite-state-machine. The upcoming graph/equivariant transformers should work a lot better in this domain.
- "Worse is better". It doesn't matter if the AI system is sometimes wrong, as long as it's cheaper. For businesses it's more cost effective to hire a low skilled human who can wrangle Codex 3.0 compared to a skilled human at FAANG salaries.
It's early days for this field. Given the current pace of progress I'm pretty certain that in 10 years almost all repetitive, non-creative programming tasks will be delegated to AI systems (eg. current "CRUD" patterns). The "real" programmers will move up the stack and become more like system architects.
With the surging popularity of no-code / co-pilot / ???, one sort of shudders to imagine the devenv of 2030. Humans taken out of the loop. Everything locked down. Natural language activated. Deep RL generated. To paraphrase Lester Bangs about the demise of rock and roll, computational liberation is in a very dangerous place right now ;)
People are still using SVN, Angular.js, Eclipse/vim/Visual Studio and all sorts of good ol' boring tech out there. Even cool kid tech like React or Golang are older than my kids.
Hashicorp stuff seems to be consistently gaining popularity though.
I actually don't see any trade offs (it's my goto framework for UI's) or why no one should be exclusively using it for app UI.
[0]: https://flutter.dev/