I think a lot of people have a sort of “slot machine” experience with it at some point. You just start firing off prompts on some new project, wait a few seconds, see what prize you got. Then you start doing that over and over just letting the LLM code and code and not even review what it’s doing. It really is like getting hooked on gambling. You’re getting a thrill from anticipation, not the actual results.
This is what I personally consider “vibe coding”, not simply using LLMs or agents or whatever in your workflow
Like I just mentioned in another comment, they're also a way to get you back to work ASAP. Just about everyone NOT working a comfy white collar office job needs to be working in order to make money. Time off is less income, people can't afford that so they do what they need to do to get back to work.
Sprained ankle? Injured back? Headache? Broken bone? All things that people work through everyday with some NSAIDs because calling out sick means losing income
Well for one thing, in America, you gotta get back to work.
Work a manual labor job or one where you're on your feet all day and sprained your ankle? Would you rather miss a week of pay (or worse lose your job) or take some pain killers and work through it?
We don't, that's why we do review it. We also do things like communicate with teammates, have expectations of not wasting other people's time, and try to uphold standards and meet SLAs. Maybe people should worry about why their teams are so dysfunctional rather than how the code was produced
I think the response would be something about the value of enjoying art and "supporting the film industry" when streaming vs what that person sees as a totally worthless, if not degrading, activity. I'm more pro-AI than anti-AI, but I keep my opinions to myself IRL currently. The economics of the situation have really tainted being interested in the technology
Not at all. Submitting untested PRs is a wildly outside of my experience. Having tests written to cover your code is a pre-requisite for having your PR reviewed on our team. "Does it work" aka passing manual testing, is literally the bare minimum before submitting a PR
It is really interesting to watch them for a while. QWEN keeps outputting some really abstract interpretations of a clock, KIMI is consistently very good, GPT5's results line up exactly with my experience with its code output (overly complex and never working correctly)
I'm guessing, in terms of angle (ie "above" or "our perspective"), this is pretty much what you'd see. We collect data sitting here on planet earth, so more or less, everything we see is from the same vantage point in the universe. The overall size and shape of things can be calculated by relative motion of objects in space. We also can make educated guesses about the shape of our own galaxy by observing others from the outside.
Write some code and then ask Claude to diff your changes and write a commit message. Now the internet hates you