The LLM is what I use to build the specific UI using the components from these great UI libraries. There’s practically no logic involved here, it’s just handling layout for components and hooking up events. It’s fantastic to be able to take a JSON payload from an endpoint throw it at a model and get a reasonable UI in seconds.

lol I think this mostly happens when people give too broad a task to llms

That said, there is a lot of boring code out there. For example, most UIs are basically just doing CRUD operations, and once you’ve written enough of these things it’s not really that exciting anymore.

Yeah, I find LLMs are really nice for learning a new language when you know what you want to do, but not the specific syntax or best patterns. I’ve also found LLMs are great for stuff like crafting SQL queries, one off shell scripts, and building UIs. They can write certain kinds of code fairly well nowadays, but you want to keep the problem scope clear and focused.

that makes it even funnier

it’s about the amount of coffee I need to wake up in the morning :)

happy now?

a good metaphor :)

That would make for a pretty heavy plugin, it still takes sizable resources to run even smaller models locally.

Indeed, and this is why the whole concept of unskilled jobs is pure nonsense.

I find this is a really common trope where people appreciate the complexity of the domain they work in, but assume every other domain is trivial by comparison.

Exactly, to eliminate the need for programmers you would need AGI, and that would simply mean the end of capitalism because at that point any job a human does can be automated.

I expect you could probably do a lot better with a model trained on a specific language, and it could probably be much smaller than a general model as well.

Could be, I’ve only tried it with js and Clojure.

I find DeepSeek is incomparably better at coding tasks

I mean sure, I can do this with Clojure too

Babashka v1.12.197 REPL.
Use :repl/quit or :repl/exit to quit the REPL.
Clojure rocks, Bash reaches.
user=> (float (+ (/ 1 10) (/ 2 10)))
user=> 0.3

no idea, but the mere fact that the question can be reasonably asked says volumes

A couple of years ago, I got laid off from a job and they were nice enough to give me two months severance. It was the happiest I can remember being in my adult life.

Kubernetes solves a problem for very large scale the kind that Google has. However, it’s complete overkill for most use cases. It’s an incredibly complex tool that takes a lot of skill to use properly and it’s completely unnecessary for most applications. As a corollary, I find there’s an obsession with aggressively applying microservice architecture nowadays which introduces a huge amount of orchestration overhead.

for sure, this is a very resilient approach that makes it incredibly difficult to take code down