What we have now are LLMs that some consider to be good at tasks that are incremental, limited in scope, and require constant human oversight with many iterations.
What you want is an LLM that is exceptionally good at completely rewriting a poorly written codebase spanning tens or hundreds of thousands of lines of code, which works reliably with minimal oversight and without introducing hundreds of critical and hard to diagnose bugs.
Not realizing that these tasks are many orders of magnitudes apart in complexity is where the "unfounded optimism" comment comes from.
What you want is an LLM that is exceptionally good at completely rewriting a poorly written codebase spanning tens or hundreds of thousands of lines of code, which works reliably with minimal oversight and without introducing hundreds of critical and hard to diagnose bugs.
Not realizing that these tasks are many orders of magnitudes apart in complexity is where the "unfounded optimism" comment comes from.