[thankschuck]

> I want this React component and that component to talk through shared context. Go. The code that comes back is often cleaner than what I would have coded myself, and it is written in a fraction of the time.

I am sorry, but if you think that your LLM instance codes BETTER than you then it means (to me) that you are bad at programming.

Faster, ok. But BETTER...

[pmontra]

I've been working on a Rails + Vue project for 3 years. Plenty of experience of Rails (all of its years) but not much experience of SPA frameworks before then. The first year has been slow on the Vue side because I had to learn its weirdnesses (among the top 5: how you have to call dispatch("file/function") instead of calling that function right away and all the contort ways of passing data up and down between component trees). Then LLMs started to help and after three years I noticed that I could handle the code made alone. I would have probably reached this point in six months. On one side LLMs are crutches that slow down learning. On the other they made me deliver software faster, at least at the beginning.

[foldr]

This really depends a lot on your level of expertise in a given domain and how much relevant training data the LLM has. I find that LLMs are often better than me at avoiding common React gotchas. This isn't really surprising, given that I write React code once in a while and the LLM has been trained on vast amounts of React code. This doesn't mean that I'm bad at programming. If I was writing lots of React code every week, I'd probably get better at it than the LLM.

[bigyabai]

So, what does your React component and context handler look like? Feel free to paste your response below, I'll only judge you relative to what AI is capable of.