GitHub wants to spam open source projects with AI slop
GitHub wants to spam open source projects with AI slop

GitHub wants to spam open source projects with AI slop

GitHub wants to spam open source projects with AI slop
GitHub wants to spam open source projects with AI slop
sexy skeleton....
They earn way to much money if they have time to create this crap.
The article says it well - this sort of thing may encourage people to get involved in open source github alternatives just as this slop on windows has moved some people to linux.
I dont believe we will get a huge movement but every little bit grows the open source community which benefits everyone. It doesnt need to be the majority it just needs to be a growing minority to thrive and grow. Ironically Microsoft has always been very good at encouraging people to get involved in open source ventures just by behaving badly.
quite amazing that (a) Microsoft basically bought open source with GitHub (b) and are now trying to fuck it up so hard people leave
And that's why I Codeberg now. I even give a few bucks every month to pretend I'm paying a subscription.
Someone should come up with a catchy alliterative name for this tactic of incorporating competing ideas in Microsoft product family, augmenting them with their own proprietary crap and pulling the rug after achieving lock-in. Maybe call it AAA for "Adopt, add on, annihilate".
I've tried a few code review tools at work and while a few suggestions are useful, most have been useless.
I made a contribution to a project that is using AI generated reviews and the AI bot gave me something like five or six suggestions, every one of which was wrong in some way or another. Only one of the suggestions was worth considering and the code that it generated to implement that suggestion was bad to the point of being baffling.
Is that Windsurf? My lot have just added that. Keeps suggesting making the path to every target in the build pipeline the same so that they'd overwrite each other, or perhaps implement the worst null-checking code I've ever seen.
The problem with suggesting 99% stupid shit is that I'm going to ignore the 1% that it identified correctly. If it limited itself to trivial syntax errors then it might have quite a useful hit rate, but we already have tools that do that.