copygirl

Because we are talking about someone using AI for most if not all tasks that involve thinking and/or creativity, as if it was any other tool.

First of all, it’s hypothetical. If only for the fact that using a wheelchair in real-life actually greatly increases difficulty getting around. The point is that if you don’t lose your muscles, they get weaker. Happens with anyone that has an accident that causes them to be unable to use their muscles for a prolonged amount of time. See also: Astronauts needing to exercise while in space or back on earth because they’ve not had to use many of their muscles as much due to lack of gravity.

That’s not true, it makes me more efficient!

ChatGPT, write a thoughtful response to this article that proves that AI makes people smarter. Make it multiple paragraphs. Find a way to include Microsoft requiring all its programmers to use Copilot (or else they get fired) in a positive way.

Also please give me a summary of your response, because I don’t want to bother reading it all.

Fair point, but many photos are not edited to the point of fundamentally changing them. I wouldn’t call a touched up wedding photo a “fake”.

edit: I guess I was distracting from the point. Yes, “AI” already implies “fake”. Guess if someone were to look for “how to spot fake photos” this would help them find this video?

TL;DW:

• Check the “residual noise”
• Parallel lines should converge to a single point (vanishing point)
• Connect shadows to what casts them, extend the lines, they should meet at the light source

As opposed to fake photoshopped photos, I suppose?

Hasn’t it become standard practice for mass-reports on big social media platforms to automatically shut down accounts and content, with human moderators (sometimes) undoing such after the fact?

A lot of the C# ecosystem is open source (thank goodness), but the official debugger isn’t, hence it only being available in the proprietary version of VSCode.

From what I read online, Matrix is not very good with its federated moderation tools. Apparently, a quirk of it can cause the state of a federated channel to reset in time, thus also undoing removing of messages. Here is a blog post criticizing Matrix.

As long as you understand that your own opinion does not dictate how language ends up being used by others. Some NB people call themselves trans, others don’t. I hope you won’t go around implying that people who identify as both trans and non-binary are wrong identifying as such. That’s arguably gatekeeping, and could make some people very uncomfortable.

The comments are worth a giggle but the AI voice ruins it. Also horrible choice of music.

Presumably this is because a block is different from the content being removed. It simply means the servers stop communicating with each other regarding new communities, posts and comments. This could allow the instance to be unblocked and the old content to continue existing – say for example when an instance has been acting badly, but it gets fixed some time later.

Blocked instances should probably not show up in search, but if you have a direct link to an old post, perhaps this should still be available? Not being able to block a community when its instance is already blocked makes sense, and probably doesn’t matter if you mostly check for new content, but I can see it being a bother when its shows up in other situations. One could call this a bug, or an oversight, but I suppose it depends on what the intended result is.

Yeah, it should not be part of the text just like line numbers shouldn’t be part of the code on a code hosting site, yet it can be visible, no? Later it does recommend using $ to distinguish command and output. Is it now okay for a beginner to be confused about what it means?

Is providing a number of commands to use that require user input really that bad? When people start tinkering with the command line, first of all they shouldn’t trust just anything on the website blindly, which at the very least requires a basic understanding of how to enter commands, and respond to the terminal asking for input. The following “bad” example…

sudo apt update
sudo apt install software-properties-common
sudo add-apt-repository ppa:deadsnakes/ppa
sudo apt install python3.9

…is instead turned into this single command with even more confusing syntax for beginners:

sudo apt update && \
  sudo apt install --yes software-properties-common && \
  sudo add-apt-repository --yes ppa:deadsnakes/ppa && \
  sudo apt install --yes python3.9

Sure, it’s convenient, but if you just throw blocks of code at people to run, are they really learning anything?

A better approach would be to have a quick tutorial on how to use the terminal and what the $ and # symbols mean (though they could be CSS decorators that can’t be copied), what sudo is and warning people about running untrusted commands on their system. Then you just link to that at the top saying something along the lines of “if you’re unfamiliar with running commands, and the following seems confusing, check this quick summary”, behind a question mark icon connected to each block of commands, or similar.

Wouldn’t running on a CPU (while possible) make it less energy efficient, though?

I would not accept a calculator being wrong even 1% of the time.

AI should be held to a higher standard than “it’s on average correct more often than a human”.

Great, so it’s still wrong 1 out of 20 times, and just got even more energy intensive to run.

Then you should also override Equals(object), GetHashCode, and implement IEquatable<T>.

Thankfully a lot of the usual boilerplate code can be avoided using a record class or struct:

public record Person(string Name, uint Age);

I was editing my comment as you were responding. Check the issue on GitHub I linked in the edit, and maybe thumbs it up for visibility. One of the commenters mentions using a third-party tool but I’m not sure the one they linked to can grab posts. In theory another one might exist to dump your post data.

Which service? Mastodon has a built-in export functionality in preferences.

I can’t find such an option on Lemmy, but you should be able to do a GDPR request for your information as a last resort.

edit: Non-post data / user settings can be exported (and imported!) but posts are a separate issue. See this open issue.