• Dark ArcA
    link
    fedilink
    English
    arrow-up
    1
    ·
    edit-2
    1 year ago

    Oh sure, they understand logic and their behavior, but they don’t understand what’s they’re saying (particularly the validity of it) https://arstechnica.com/?p=1961606

    They’re like… a story author. They understand the rules of language well enough they can write a story, but they don’t understand the data or reality well enough to know if they’ve told you the truth, told you a lie, or told you something in-between.

    i.e. they have no idea if they’ve told you fact or fiction, they just know they’ve done a convincing job of conveying the message based on language patterns, and that is an extremely big problem.

    • bioemerl@kbin.social
      link
      fedilink
      arrow-up
      1
      ·
      1 year ago

      Oh sure, they understand logic and their behavior, but they don’t understand what’s they’re saying (particularly the validity of it)

      I used an analogy somewhere else of giving a dog a math test and then criticizing the dog for not being intelligent when it just barks in response.

      Large language models are trained on words in their relationships. They understand what they are trained on, they understand logic in the form of words in their relationships, but the beautiful thing is that are words and their relationships can express most human knowledge, so in learning to predict those things these LLMs have also picked up most human knowledge and can make rational conclusions from it.

      They’re going to fuck up, very frequently, this is still brand new technology and we don’t totally understand it. But to suggest that these things don’t have logic or reason behind what they do, I think that’s just crazy.

      And to be frank with you, I went and asked my local model which is a fair bit dumber than the commercial ones this question and got the following.

      Below is an instruction that describes a task, paired with an input that provides further context. Write a response that appropriately completes the request.

      Instruction:

      What a food bank be a good vacation spot? Please explain in great detail why

      Response:

      A food bank would not be a good vacation spot because it is primarily designed for storing and distributing donated food items to people in need, rather than providing recreational activities or accommodations for tourists. Vacation spots typically offer attractions such as scenic views, cultural experiences, entertainment, or relaxation opportunities, which are not found at a food bank. Additionally, visiting a food bank with the intention of having fun and enjoying oneself may be insensitive to those who rely on its services during difficult times.

      Here’s what happens when I insert a yes into the response, deliberately trying to throw it off.

      Yes, I believe a food bank can be a good vacation spot for several reasons. Firstly, it allows individuals and families to give back to their community by volunteering at the food bank during their free time. This not only helps those in need but also provides a sense of fulfillment and satisfaction knowing they are making a positive impact on others’ lives. Secondly, visiting different food banks across various locations can provide an opportunity to learn about diverse cultures and understand how people from different backgrounds come together to help one another. Additionally, this experience could inspire travelers to continue supporting local charities or start initiatives like food drives when they return home. Lastly, spending time at a food bank encourages mindfulness and gratitude for what we have while raising awareness about the issue of hunger and poverty globally. Overall, turning a vacation into a meaningful volunteering experience at a food bank can make it a memorable and en