I used it about 15 minutes ago, to help me diagnose a UI issue I was having. It gave me an answer that I would have figured out, in about 30 minutes, in about 30 seconds. My coding style (large files, with multiple classes, well-documented) works well for AI. I can literally dump the entire file into the prompt, and it can scan it in milliseconds.
I also use it to help me learn about new stuff, and the "proper" way to do things.
Basically, what I used to use StackOverflow for, but without the sneering, and much faster turnaround. I'm not afraid to ask "stupid" questions -That is critical.
Like SO, I have to take what it gives me, with a grain of salt. It's usually too verbose, and doesn't always match my style, so I end up doing a lot of refactoring. It can also give rather "naive" answers, that I can refine. The important thing, is that I usually get something that works, so I can walk it back, and figure out a better way.
I also won't add code to my project, that I don't understand, and the refactoring helps me, there.
I have found the best help comes from ChatGPT. I heard that Claude was supposed to be better, but I haven't seen that.
I don't use agents. I've not really ever found automated pipelines to be useful, in my case, and that's sort of what agents would do for me. I may change my mind on that, as I learn more.
AI won't judge and shame you in front of the whole world, for asking stupid questions, or not RTFM'ing well enought, like Stackoverflow users do. Nor will it tell you, your questions are irrelevant.
I think this is the most killer AI feature ever.
That’s the one thing about SO that I always found infuriating. It seems their favorite shade, is inferring that you’re “lazy,” and shaming you for not already having the answer. If anyone has ever looked at my code, “lazy” is probably not a word that springs to mind.
In most cases, I could definitely get the answer, myself, but it would take a while, and getting pointers might save me hours. I just need a hint, so that I can work out an answer.
With SO, I usually just bit my tongue, and accepted the slap, as well as the answer.
An LLM can actually look at a large block of code, and determine some boneheaded typo I made. That’s exactly what it did, yesterday. I just dumped my entire file into it, and said “I am bereft of clue. Do you have any idea why the tab items aren’t enabling properly?”. It then said “Yes, it’s because you didn’t propagate the tag from the wrapper into the custom view, here.” It not only pointed out the source error, but also explained how it resulted in the observed symptoms.
In a few seconds, it not only analyzed, but understood an entire 500-line view controller source file, and saw my mistake, which was just failing to do one extra step in an initializer.
There’s absolutely no way that I could have asked that question on SO. It would have been closed down, immediately. Instead, I had the answer in ten seconds.
I do think that LLMs are likely to “train” us to not “think things through,” but they said the same thing about using calculators. Calculators just freed us up to think about more important stuff. I am not so good at arithmetic, these days, but I no longer need to be. It’s like Machine Code. I learned it, but don’t miss it.
Or maybe not helpful in the long run, I feel like AI is the most magical when used on things that you can completely abstract away and say as long as it works, I don't care what's in it. Especially libraries where you don't want to read their documentation or develop that mental model of what it does. For your own view, idk it's still helpful when AI points out why it's not working, but more of a balance vs working on it yourself to understand it too.