I am using it in exactly the same way. I have access to ideas and code that I never would have tried before.
These other well-documented experiences that sound super perfect don’t match mine, though I’m a fan! Here’s how it goes for me.
I am a premium user and all, but it still bails on me every single time it’s going well, by writing half of a function or saying “network error”, then losing context.
I have to say “you stopped generating that last part, please continue and don’t bother qualifying.” Then it wastes more tokens apologizing anyway.
Whoever trained these models to apologize and say “as a large language model, I can’t promise that I know how Bjarne Stroustrup would refactor that bitshift, and yada yada” really wasted a lot of our productivity. And saved no one from any dangers of AI.
These other well-documented experiences that sound super perfect don’t match mine, though I’m a fan! Here’s how it goes for me.
I am a premium user and all, but it still bails on me every single time it’s going well, by writing half of a function or saying “network error”, then losing context.
I have to say “you stopped generating that last part, please continue and don’t bother qualifying.” Then it wastes more tokens apologizing anyway.
Whoever trained these models to apologize and say “as a large language model, I can’t promise that I know how Bjarne Stroustrup would refactor that bitshift, and yada yada” really wasted a lot of our productivity. And saved no one from any dangers of AI.