I see three groups of developers who like LLMs:
1) Non-programmers, or programmers who always found coding a drudgery and didn't care for improving in that dept. For them LLMs are obviously a godsend.
2) Programmers who didn't have that issue and could write code fast, but weren't big on correctness. These are the people who hate writing tests. Since they don't care much about correctness, they don't worry about the slop much either - if it works, ship it!
(Part of the hype is also because managers see most programmers falling into #1 or #2 and are cool with #2s mindset. )
3) Programmers who understand LLMs shortcomings but they just want to churn out stuff they previously had no time for and for those, correctness or maintainability is not high on their list.
So you are not going to convince these people. They are getting real value.
The problem is, EVERYONE is slowly being forced to adopt LLMs and work this way, and the notion of outcome over output we have been trying to spread is disappearing.




