For open source programs, AI coding tools are a mixed blessing

Honestly? The problem isn't that AI tools write bad code—it's that they write *plausible* code that someone else has to maintain in 2027. You can scaffold a feature in 45 minutes that'll take 6 weeks to patch when it breaks in production. The real tax isn't the build time, it's the pile of unmaintained cruft that nobody signed up to support. For open source especially, you're recruiting volunteers to debug some junior dev's LLM output. That's not a feature gain, that's a hostage situation.