I’m not sure what their simple function was but I tried to use Claude to recreate C++ code to implement the algorithms in this paper as practice for me in LLM use and it didn’t go well. But I’ll be the first to admit that I’m probably holding it wrong.

https://users.cs.duke.edu/~reif/paper/chen/graph/graph.pdf

Can't remember, but it was something very basic - a 10/15-line routine that a first-year student would write in 3 minutes if they knew the relevant API. The reason I asked the model in the first place is because I didn't know the API. If memory serves, the model inverted an if or a loop condition.

Did you use one of the latest frontier reasoning models? If not, how is your experience relevant?

In what world is this an appropriate thing to say to someone?

In the world where you do not claim that LLMs suck today based on your attempt to use some shitty model three years ago.

In the creed of "AI is perfect, if you claim otherwise you're broken" that so many here embrace.

So you tried it once and then gave up?

I didn't give up, I just know that I can only use a model when I have the patience to work with something I can't trust at all on anything. So that's what I do.

Sounds like the spirit of my question remains intact