Actually, most VPN providers explicitly label the virtual locations as such, I think the famous ones at least do it (ex: Proton and NordVPN even explain them in their respective docs).
It depends on whether the VPN is lying to you. Proton, for example, makes them quite explicit in the software and even lists them for you here: https://protonvpn.com/support/how-smart-routing-works and seems like NordVPN also has a page explaining that.
I have this impression that LLMs are so complicated and entangled (in comparison to previous machine learning models) that they’re just too difficult to tune all around.
What I mean is, it seems they try to tune them to a few certain things, that will make them worse on a thousand other things they’re not paying attention to.
Anything that is very specific has the same problem, because LLMs can’t have the same representation of all topics in the training. It doesn’t have to be too niche, just specific enough for it to start to fabricate it.
One of these days I had a doubt about something related to how pointers work in Swift and I tried discussing with ChatGPT (don’t remember exactly what, but it was purely intellectual curiosity). It gave me a lot of explanations that seemed correct, but being skeptical and started pushing it for ways to confirm what it was saying and eventually realized it was all bullshit.
This kind of thing makes me basically wary of using LLMs for anything that isn’t brainstorming, because anything that requires knowing information that isn’t easily/plentifully found online will likely be incorrect or have sprinkles of incorrect all over the explanations.
It doesn’t really solve it as a slight shift in the prompt can have totally unpredictable results anyway. And if your prompt is always exactly the same, you’d just cache it and bypass the LLM anyway.
What would really be useful is a very similar prompt should always give a very very similar result.
This doesn't work with the current architecture, because we have to introduce some element of stochastic noise into the generation or else they're not "creatively" generative.
Your brain doesn't have this problem because the noise is already present. You, as an actual thinking being, are able to override the noise and say "no, this is false." An LLM doesn't have that capability.
I'm so thankful for nixos for making it hard for me to give in to that temptation. you always think "oh just this once". but with nixos I either have to do it right or not bother.
$ ./Downloads/tmp/xpack-riscv-none-elf-gcc-15.2.0-1/bin/riscv-none-elf-cpp
Could not start dynamically linked executable: ./Downloads/tmp/xpack-riscv-none-elf-gcc-15.2.0-1/bin/riscv-none-elf-cpp
NixOS cannot run dynamically linked executables intended for generic
linux environments out of the box. For more information, see:
https://nix.dev/permalink/stub-ld
You have to go out of your way to make something like that run in an fhs env. By that point, you've had enough time to think, even with ADHD.
It sort of does actually, at least if you don't have nix-ld enabled. A lot of programs simply won't start if they're not static-linked, and so a lot of the time if you download a third-party script, or try to install it when the `curl somesite.blah | sh`, it actually will not work. Moreover, it also is likely that it won't be properly linked in your path unless you do it thr right way.
Maybe they can with postinstall scripts, but they usually don't.
For the most part, installing packaged software simply extracts an archive to the filesystem, and you can uninstall using the standard method (apt remove, uv tool remove, ...).
Scripts are way less standardized. In this case it's not an argument about security, but about convenience and not messing up your system.
