I haven't had the same experience. Here are some of the significant issues when using o1 or claude 3.7 with vscode copilot:
* Very wreckless in pulling in third party libraries - often pulling in older versions including packages that trigger vulnerability warnings in package managers like npm. Imagine a student or junior developer falling into this trap.
* Very wreckless around data security. For example in an established project it re-configured sqlite3 (python lib) to disable checks for concurrent write liabilities in sqlite. This would corrupt data in a variety of scenarios.
* It sometimes is very slow to apply minor edits, taking about 2 - 5 minutes to output its changes. I've noticed when it takes this long it also usually breaks the file in subtle ways, including attaching random characters to a string literal which I very much did not want to change.
* Very bad when working with concurrency. While this is a hard thing in general, introducing subtle concurrency bugs into a codebase is not good.
* By far is the false sense of security it gives you. Its close enough to being right that a constant incentive exists to just yeet the code completions without diligent review. This is really really concerning as many organizations will yeet this, as I imagine executives are currently the world over.
Honestly I think a lot of people are captured by a small sample size of initial impressions, and while I believe you in that you've found value for use cases - in aggregate I think it is a honeymoon phase that wears off with every-day use.
I've been using it daily for years. Mostly asking questions in a separate chat window/app and then working its response into my code. And then I sped up the feedback loop when I migrated to Cursor where I began pushing the envelop and asking it to do more.
I think what wears off is that we're less impressed and then we start demanding more and more from it and getting frustrating when it can't do it. But that's different than a honeymoon phase wearing off. It's like how we're not really impressed by image gen anymore, we expect it.
But as an example of a selfish sense of loss I've experienced, I used to pride myself in being the only developer on any team who ever learned CSS. I could architect a good grid/flex layout with a lot of thought. I could do little things like make text in a small UI component truncate into {3 letters} + ellipses when its parent was too small. And most of all I could polish UIs to a point where I'd say they were perfect, even a form.
Now, LLMs are really good at doing the mechanical parts of the things I spent so much time learning. Like I originally said, I'm not shedding tears over here saying it's so unfair. But there is a sense of loss. And when I figured most people reading my comment would misinterpret this, I removed my comment. Because you can't make descriptive claims about how you feel online, it can only be interpreted as a normative value judgement about the world. Because I guess that's what it is 99.9% of the time someone expresses a feeling they feel, but not in this case.
Finally, the right way to see it is that now I can polish the UI to perfection, but I don't need to be a CSS expert anymore. Nobody needs to be. You can get an idea of how you want the UI to work and ask the LLM "make this one bit of text be the one that truncates if the window is too narrow" and it does it. And that's fkin magic.
Have you tried copying the compilation errors back into the prompt? In my experience eventually the result is correct. If not then I shrink the surface area that the model is touching and try again.
yes ofcourse. it then proceeds to agree that what it told me was indeed stupid and proceeds to give me something even worse.
I would love to see a video of ppl using this in real projects ( even if its open source) . I am tried of ppl claiming moon and stars after trying it on toy projects.
Yeah that's what happens. It can recreate anything it's been trained on - which is a lot - but you'll definitely fall into these "Oh, I see the issue now" loops when doing anything not in the training set.
I regularly make knee-jerk comments on HN that I delete a minute later. Something therapeutic about it.
My comment isn't one I wanted on my "record". You responded to it and I saw your response before deleting my comment. What's the harm? It's obvious I removed my comment.
> I regularly make knee-jerk comments on HN that I delete a minute later. Something therapeutic about it.
I'm really curious about this. Doesn't it feel selfish to you to subject the public to your internal anxieties? It's the same reason I don't unload on everyone around me.
EDIT: I'm not trying to dunk on you. You're being honest so thanks for that.
