Making
Points and comments are a snapshot, not live.
The author argues that AI-generated output lacks the personal fulfillment of making something oneself.
Beej Jorgensen, a Gen-X hacker and CS instructor, contrasts the satisfaction of hand-coding a 177-line flash card app with the hollowness of AI-generated work. He distinguishes 'making' from 'asking to be made,' comparing AI prompting to managing a contractor rather than using a tool like a compiler. While AI can produce results faster, the loss of craft and personal agency in creation leaves him feeling he did nothing. He acknowledges gray areas like compilers but maintains that AI's human-like interaction shifts the relationship from maker to requester.
What commenters are saying
Commenters split on whether LLMs diminish or enhance personal satisfaction in building. One camp argues that the loss of agency and the inability to reason about input-output relationships makes AI-created work feel hollow, comparing it to steroids in bodybuilding. Another finds LLMs liberating for side projects that would otherwise languish due to time constraints, enabling them to build things they actually want to use. A key observation is that predictability and the ability to debug via source code are what separate compilers (trusted tools) from LLMs (black-box collaborators).