• 0 Posts
  • 47 Comments
Joined 2 years ago
cake
Cake day: June 18th, 2023

help-circle












  • I concede there was a window of time where that was true. Too early, and AI had nothing to do with the current methods and tried a more symbolic approach. Too late (i.e. now) and the term has been pushed so hard by corpos (because of the superficial semantic reading “AI” => “intelligence” for marketing hype) and it means for a lot of people, whether you like it or not, “ChatGPT”, and it alienated them.

    Unproductive? Probably (and for sure in the context of your work). Pointless? Absolutely disagree when companies force that vision of intelligence on us. This has a social impact.

    Now maybe this is pointless to you, and that’s ok, but it’s not to a lot of other people.



  • That is mostly because, at the time, in the reduced context of videogames, it was clear that AI was used to describe the behaviour of non playable entities.

    And also, I don’t think video game terms were ever taken to have scientific accuracy (at least I hope not) or, more importantly, ever tried to imply that these entities exhibited “intelligence”.

    Now an entire subfield of statistics is being called AI by virtue of the fact that we often do linguistic abuses when it comes to talking about computers or code (something that Dijkstra was vehemently against in this fantastic note about teaching compsci). I don’t know why statisticians felt the need to hype up gradient descent by calling it “learning” but here we are.

    Now I know I am caricaturing, but the point I am trying to make is that, now that the cat is out of the bag, and that “AI” is not just an academic term but has been willfully used to get money and to sell products with anything and everything, the unfortunate effect is that for a lot of people, AI = LLMs mostly. And I’d say amongst these people, a number, me included, would like it to stop using that term entirely because of that abuse AND because of the suggestion that it exhibits intelligence, in that context.

    I get that it sucks for you and you feel attacked if you do anything that has to do with machine learning or deep learning, but again, context is important, and this comm is pretty clearly against the slop generators (and the term AI altogether for the reasons mentioned), not necessarily all modern tools of statistical analysis and pattern recognition.


  • EDIT: Just saw that Malik already did mention this more succinctly. Please feel free to ignore me.

    ORIGINAL COMMENT: The comments here already cover a good bit, esp. with the link to Piotr’s blog post.

    However I don’t see anyone reacting to your mention of the snap store.

    If you want some details about that, you can read here: https://linuxmint-user-guide.readthedocs.io/en/latest/snap.html

    But in a few words, distributing software is kinda of a mess in Linux at first glance, for various technical reasons.

    To caricature, you used to only install the packages from your distribution (mint for you) repositories, and if a program wasn’t in it, you had to either compile it or jump through other hoops.

    Then came other formats which made distributing software across Linux distros easier, with some caveats. Two notable ones are Snap and Flatpak.

    Snap was made by the guys behind Ubuntu and mint is an offshoot of Ubuntu that made the willful decision to not do snaps by default after a number of fiascos.

    My advice would be: try installing software through the normal mint repositories, ideally the non Flatpak version. If it does not exist there or is buggy or whatever, consider the Flatpak. Only failing that should you look into snap IMO.