I think it's helpful to separate the problems of the technology itself from all the BS that big tech companies engage in when developing and pushing it.

Say you only use small, open-weight models that are hosted locally in a data center that is conscientious, environmentally clean, has strict data privacy policies and doesn't pull the bullshit big-tech do.

You use them minimally and responsibly, you don't over-rely on them. Is there a problem in that case?

Also, a ban without control isn't a ban. And making sure that not a single line of code was writen with the help of some AI seems more than complicated... is it just possible?

And how much working time would it take to audit all packages to make sure it is 100% wrote by some human?

It seems more realistic (and more honnest) to allow the part that would not be found by a reasonable audit of the code.

For all those that say 'AI causes global warming' - you're correct, but unless you've got your own solar panels, so does your non-AI gaming rig. And your gas-driven car.

so fucking what? stop trying unless you're perfect? every bit helps! fuck you

Agreed. I think they should have pushed for stronger disclosure requirements, but I also realise that isn't actually enforceable.

I also find it very telling that people will talk about the negative parts of AI, but not acknowledge other automation tools that perpetuate the same problems. My experience is that most people's complaints have a real basis, but they are just mis-diagnosed.

The problems seem to be our economic and societal structures, and AI is the new fancy tool that exacerbates those problems. But I've not heard a good reason why AI itself is that problem.

I disagree. You are an AI evangelist and quite closed-minded. Having a "balanced" approach, if there was one, would be to label any AI usage explicitly. If you're afraid to label your AI usage, you're a fucking coward.

But, damn, let's not blame the billion dollar data centers for global warming, guys! It's the individual decisions you make every day that kills the planet! If you believe this, the AI has fried your brain. It's not one authoritarian government, it's a different one! Actual paste-eating behavior. My work has improved with LLM usage! A common saying from people who were not good at their job, and are somehow now worse without them realizing it.

Any decent developer I have found has hated LLMs because they do not work anywhere near as well as people who are shit at things think they do. Personally, I don't really like smashing the Skinner Box button hoping the output works this time because I like programming, and having Claude rape my codebase seems like a bad idea to me if I ever want to go back in there. But, then again, I like knowing how my code works, so maybe I should just lobotomize myself and I'll understand why you like it so much.