I’ve always been the kind of developer that aims to have more red lines than green ones in my diffs

> Unseen were all the sleepless nights we experienced from untested sql queries and regexes and misconfigurations he had pushed in his effort to look good. It always came back to a lack of testing edge cases and an eagerness to ship. If you do this you are creating a rod for your own back: You need management to see the failures & the time it takes to fix them, otherwise they will assume everything is fine & wonderful with their new toy & proceed with their plan to inflict it on everyone, oblivious to the true costs + benefits.

I think there will still be room for "debugging AI slop-code" and "performance-turning AI slop-code" and "cranking up the strictness of the linter (or type-checker for dynamically-typed languages) to chase out silly bugs" , not to mention the need for better languages / runtime that give better guarantees about correctness. It's the front-end of the hype cycle. The tech-debt problems will come home to roost in a year or two.

You have to go lower down the stack. Don't use AI but write the AI. For the foreseeable future there is a lot of opportunity to make the AI faster. I am sure assembly programmers were horrified at the code the first C compilers produced. And I personally am horrified by the inefficiency of python compared to the C++ code I used to write. We always have traded faster development for inefficiency.

If the company values that 10x speedup, there is absolutely still a place for you in this environment. Only now it's going to take five days instead of two, because it's going to be harder to track that down in the less-well-structured stuff that AI produces.

AI can definitely produce a deletion. In fact, I commonly use AI to do this. Copy some code and prompt the AI to make the code simpler or more concise. The output will usually be fewer lines of code. Unless you meant that AI won’t remove entire features from the code. But AI can do that too if you prompt it to. I think the bigger issue is that companies don’t put enough value on removing things and only focus on adding new features. That’s not a problem with AI though.

AI can do deletions and refactors, and 10x speedups. You just need to push the latest models constantly.

>AI use makes the metric for success speed-to-production Wasn't it like that always for most companies? Get to market fast, add features fast, sell them, add more features?

AI writes my unit tests. I clean them up a bit to ensure I've gone over every line of code. But it is nice to speed through the boring parts, and without bringing declarative constructs into play (imperative coding is how most of us think).