really really really depends on what sort of code is being 'written'. 20 years ago, IDEs would automatically create boilerplate getters and setters. In large

really really really depends on what sort of code is being 'written'. 20 years ago, IDEs would automatically create boilerplate getters and setters. In large projects that's a non-trivial amount of code. IDEs can autocomplete stuff already. For most of the folks I know doing non-trivial projects, AI tools are... useful autocompletes, but not much more. So... 25% of your code was done by AI but is it the hard nitty gritty stuff? The value prop of your whole company? Or is it just lots of boilerplate that is necessary because of all the abstractions we have at our disposal today (or... all the abstractions that are required to do anything 'modern' to use a negative light on it)?