Well, people are people. It might be unpleasant and rude to systemically browse people, especially if you need to interrupt and clarify your prompts, basically treating them like chatbots.

The way I see it, you should differiante how you approach human and inhuman sources. Information from people have a lot of advantages, too. You might get your target information quickly without any "bloat" from e.g. an encyclopedia article. However you might lose a lot of key information. A person forces you into an interactive bi-directional conversation. They will get information from you and more likely add additional information you need.

For example you might get the commonly accepted translation of an ancient poem from an article. A human can give you that, too, but if they notice that you are absolutely not familiar with the subject, they might also clarify that a literal translation is not possible, that you need the context of this historical event and so on. In a inhuman source you might have skipped that information. This is how some people may use a trusted source, but still leave with fake news, because the extracted information is incomplete.

In reality you can't reduce AI to replace only "Level 1 coding" and do only "typing". It will make assumptions about these "Level 2 and 3" decisions in its generated code. To reduce or control it you have to invest more into documentation/instructions and code review. You basically change the focus based on the assumption that "Level 1 coding" with all of that "hand-crafted" code was such a big waste of time and money. But it's a made-up problem.

On top of that a lot of vibe-coded projects that appear here and there seem to not even intend to let the AI do only the typing. They don't just let the AI translate "flow" and "architecture" into code. They make the AI translate their demands into code.

The beginning confusion was an intentional trap by the author. The author's real confusion comes only later.

It's kinda weird to see coding with AI used with languages and tools that are (beginner-)developer friendly and highly abstracted on their own. It feels like a waste of abstractions that were meant for humans? Like creating a robot to help in the kitchen, but using it to call a pizza.

Neither I guess? I hate the expression "separating the work from the artist". It assumes that you can only completely ignore the context or reduce everything to the context. Humans are capable of more than that. You don't even have to actively decide how much you value the context and whether or not you are comfortable with a piece of art. It'll happen automatically.

Regarding not giving support to an artist whose views you strongly oppose, then it's about the product and not the work. There's a huge difference between opposing the power an artist gains from their product and opposing the art itself.

Could you elaborate how the configuration might not be respected? Do you mean that you've often encountered applications that save files to hard-coded paths and do not even let you change the destination path?

If you ask me, that's just bad software design. If the software is open-source, there is the option to request the developers to read the actual path of the respective well know directory from the XDG environment variables or allow configuration.

If we look at only the Blender Development Fund, the list of the corporate patrons include Meta, Google, Nvidia and even Adobe. In general, this is the situation for a lot of FOSS projects. If Anthropic as a Blender patron is bad news, then things were never good in the first place.

In my opinion, whether it's good or bad, it is a compromise that we rely on.

The user does decide, XDG user directories are optional and configurable. Since they are already established, user-friendly distros / desktop environments already pre-install them.

And what speaks against just using a new directory within your home directory as your "specific place that is user owned that isn't filed with cruft and configuration files"?

They clearly meant the cumulative age when two of the four children are holding the controllers together (must be a new Gen Alpha & Beta activity). The youngest child is 2 years old, the twins are 3 years old and the big boss is already 5 years old. There is no other reasonable way to interpret this.

Websites that let me scroll down and up without any annoying animated hindrance have my thanks.

Uh, I think it's the exact other way around. In practice most of the activity already is centralized in more general "melting pot" communities and the lack of engagement is the reason why the content is not distributed across the more specific communities.

Why is this situation not intentionally desirable (on paper)? Well, it kinda misses the whole point of the federation. Lemmy, despite decentralization, is currently more dependent on a few of its communities than the evil corpo social media. Then again, this just proves that technical centralization has always been a lesser issue with the traditional social media services and that activity is where activity is.

I still don't like the idea of one big general community. I'm certain that a lot of the people here don't want content just for the sake of content. Being forced to manually filter out most of the content would be a hot mess. On top of that, while activity might increase slightly in quantity, the quality would become even more superficial and shallow. For me personally, it'd be a reason to stop using Lemmy.

In general there is the !conlangs@mander.xyz community.