Post #2460220

2026-05-12 20:21 UTC

Anecdotally, from talking to a lot of people in person at Dutch Clojure Days, my impression is that, at least in this community, the number of people who practice or believe in full on vibe coding is very small. The number who categorically don't engage with AI is also very small. The vast majority seems to use some LLM tools for coding at least semi-regularly, and think it's useful at least for some things (not always agreeing on which things). There's huge variation though in how much and in what way it's being used. Still, the general sentiment seems to be mostly negative. There are ethical and environmental concerns, worries about jobs security, but also just a sense that it's unsatisfying to work with, with frustrating, mediocre, and inconsistent results. People who have to use it at work expressed that code review is now the bottleneck, and that they can see their codebase deteriorating week by week, with more duplication and inconsistencies. This is all very anecdotal, so make of it what you will.

Replies (3)

@otfrom@functional.cafe 2026-05-12 21:11

@plexus@toot.cat as I'm about to start on a new project those experience reports really help me. (a code base that gets worse over time doesn't sound like a good bet to me)

Open ##2837028
@rgm@cosocial.ca 2026-05-12 21:14

@plexus@toot.cat I’m trying on this fresh metaphor about our age: imagine that an intellectual chainsaw just got invented, and it’s just hard over here, trying to be at all thoughtful/critically-engaged about it while watching people who 100% do in fact know better get the intellectual clearcutting underway in quite a number of the intellectual forests I care about.

Open ##2837029
@weavejester@mastodon.social 2026-05-12 23:19

@plexus@toot.cat My experience is that LLMs are most useful for tasks that are trivially verifiable. I tend to use them for small, one off scripts or for searching through documentation. I haven't found them that useful for debugging, except as a rubber duck.

Open ##2837030