Sorry about that.

  • 0 Posts
  • 137 Comments
Joined 1 year ago
cake
Cake day: June 12th, 2023

help-circle










  • Are you saying that traditional food delivery drivers get trained specifically not to hit on people when they deliver food? I don’t have any data but I feel like that’s not really a thing. Maybe my concept of the training a good delivery driver gets is way off the mark?

    I’m also pretty sure that it’s easier to give a bad review that others will see via one of these food delivery apps than it is if you go directly to the business.

    I think we all agree that this is inappropriate and should not be happening, I just don’t see how it doesn’t apply at least equally to traditional delivery drivers.


  • I can’t say I fully understand how LLMs work (can’t anyone??) but I know a little and your comment doesn’t seem to understand how they use training data. They don’t use their training data to “memorize” sentences, they use it as an example (among billions) of how language works. It’s still just an analogy, but it really is pretty close to LLMs “learning” a language by seeing it used over and over. Keeping in mind that we’re still in an analogy, it isn’t considered “derivative” when someone learns a language from examples of that language and then goes on to write a poem in that language.

    Copyright doesn’t even apply, except perhaps on extremely fringe cases. If a journalist put their article up online for general consumption, then it doesn’t violate copyright to use that work as a way to train a LLM on what the language looks like when used properly. There is no aspect of copyright law that covers this, but I don’t see why it would be any different than the human equivalent. Would you really back up the NYT if they claimed that using their articles to learn English was in violation of their copyright? Do people need to attribute where they learned a new word or strengthened their understanding of a language if they answer a question using that word? Does that even make sense?

    Here is a link to a high level primer to help understand how LLMs work: https://www.understandingai.org/p/large-language-models-explained-with



  • Check out VanillaOS. I think it’s pretty neat. Their webpage doesn’t really get into the benefits as much as I think they should, but a very quick summary is that it leverages distrobox and some custom package manager to allow you to seamlessly install and run packages from other distros. It’s also kind of an immutable OS (but not really). It lets you pick which types of apps you want during the install (snaps, fltapak, AppImage, etc)

    I am not super in the loop about why people are so against snaps, but I don’t like the centralized nature of them, and if that’s also the general concern, then flatpak should be fine, since it’s decentralized.

    I saw a couple youtube videos about VanillaOS; I could certainly find you one of them if you want to know more.





  • We don’t even know how they arrive at the output they arrive at, and it takes lengthy research just to find out how, say, an LLM picks the next word in an arbitrarily chosen sentence fragment. And that’s for the simpler models! (Like GPT-2)

    That’s pretty crazy when you think about it.

    So, I don’t think it’s fair to suggest they’re just “a new type of app”. I’m not sure what “revolutionary” really means but the technology behind the generative AI is certainly going to be applied elsewhere.



  • I’m not sure your second point is as strong as you believe it to be. Do you have a specific example in mind? I think most vehicle problems that would require an emergency responder will have easy access to a tow service to deal with the car with or without a human being involved. It’s not like just because a human is there that the problem is more easily solved. For minor-to-moderate accidents that just require a police report, things might get messy but that’s an issue with the law, not necessarily something inherently wrong with the concept of self driving vehicles.

    Also, your first point is on shaky ground, I think. I don’t know why the metric is accidents with fatalities, but since that’s what you used, what do you think having fewer humans involved does to the chance of killing a human?

    I’m all for numbers being crunched, and to be clear (as you were, I think) the numbers are the real deciding metrics here, not thought experiments.

    And I think it’s 100% true that autonomous transportation doesn’t have to be perfect, just better than humans. Not that you disagree with this, but it is probably what people are thinking when they say “humans do this too”.