Author argues prompting Claude is programming

The author argues that prompting artificial intelligence models successfully is fundamentally a form of programming, requiring the same mental decomposition, explicit specification, and debugging loops. Software engineers excel at this because they have been trained by compilers to be precise, whereas non-programmers often struggle by treating LLMs like human interlocutors. Consequently, the author suggests that the most effective AI training for knowledge workers might be basic programming rather than learning superficial prompt templates. Open article →