pull down to refresh

101 sats \ 1 reply \ @SimpleStacker 21h \ on: Add question about AI usage to PR template by ekzyis · Pull Request #2507 devs
It might be helpful to have some kind of rubric for levels of AI assistance. Off the top of my head it could be like
  1. No AI assistance
  2. AI was consulted for help, but did not write any of the code
  3. AI provided tab-autocompletions while coding
  4. Significant sections were based on AI first drafts
  5. Vibe coded (entire first draft written by AI)
write
preview
0 sats \ 0 replies \ @adlai 1h
  1. Vibe coded (entire first draft written by AI)
My understanding1 is that true "vibe coding"2 would be level six or even seven3; using a first draft from some LLM and then doing the rest of the work yourself is almost sane in comparison to thinking that you can keep on being a serial entrepeneur without ever befriending, or at least touching the shoulders of, your keyboard jockey.

Footnotes

  1. mosly from memes, social media gossip, and a few conversations; I haven't actually followed any of the vibe coding gurus myself. I realise there are serious folks with a strong technological background who are working on making the following footnote a feasible technique... ↩
  2. true "vibe coding" is where you only talk, by voice rather than text chat; and you complain like you would to customer support when the app you're building doesn't work [or even compile]. ↩
  3. so if "true vibe coding" is level seven, then level six would be where you are continuously using AI to drive your development environment, however you still audit actual the source code; as opposed to level seven, where code reviewers of the PR might be the first humans to see the patches and resulting code. ↩
reply