@regehr I have never been quite as depressed as now, seeing a lifetime of skills rapidly depreciating to pointlessness.
@regehr I mean, sure, here's some confident developers laughing at him. And perhaps he'll turn out to be wrong.
But these models are already a lot better than I ever thought they'd be, and I don't feel confident predicting where they'll be in a year.
There is one important question that keeps getting overlooked in all the talk about LLM code generators replacing human developers, and that is "Where's the software?"
Where _is_ the software? If LLMs are actually useful and yields a significant productivity boost in software development, where's all the new software? Where are the more rapid release cycles, the more frequent updates, the long-requested features finally being added? Why has nothing changed on _that_ end?
@skjeggtroll @datarama @regehr
https://mikelovesrobots.substack.com/p/wheres-the-shovelware-why-ai-coding
That post asks the same question. The author then looks at things like the number of android apps added to the play store per month or the number of new GitHub repos. There is no real increase there.
@sebastian that mikelovesrobots also says this, which I think sums up the situation nicely:
> This whole thing is bullshit.
