The line gets blurrier the more auto-complete you use.

Agentic programming is at the end of the day a higher level auto complete, with extremely fuzzy matching on English.

But when you write a block and you let copilot complete 3, 4, 5 statements. Are you really writing the code?