You're missing the part where building a modern website is a huge amount of dev time for largely UI work. Also modern deployment is 100x more complicated than in Brook's day. I'd say 90% of my projects are on these two parts which really shows how productivity has gone down (and AI can fix)
This is mostly self inflicted though. We create complex deployments with the promise that the incremental savings will overtake the upfront costs when they rarely do (and the hidden complexity costs).
So it seems AI will just let us stretch further and make more accidentally complex systems.
The value of automation ("complex deployments") is not only incremental cost savings (ie because you don't need to do the work over and over), but also the reduction or outright elimination of human error, which especially in the case of security-sensitive activities like deploying software on the Internet can be orders of magnitude more costly than the time it takes to automate it.
Modern development is more complex, not more complicated. We’re still using the same categories of tools. What’s changed is the tower of abstraction we put between ourselves and the problem.
Back when IE was king and IE6 was still 10% of users, I did frontend web work. I remember sitting next to our designer with multiple browsers open playing with pixel offsets to get the design as close as practically possible to the mockups for most users and good enough for every one else. This isn't something LLMs do without a model of what looks good.
My current job involves exactly this (thanks not on IE) and AI is, as you said, absolutely bad at it.
And I’m saying this as someone who kind of adopted AI pretty early for code and who learned how to prompt it.
The best way to make AI worth your time is to make it work towards a predictable output. TDD is really good for this : you write your test cases and you make the AI do the work.
But when you want a visual result ? It will have no feedback of any clue, will always answer "Ok, I solved this" while making things worse. Even if the model is visual, giving it screenshots as feedback is useless too.
