4 ms·

There is no universe where an LLM helps one learn to read. You need to be able to read first to use one, and worse yet, you need to be able to think critically

• tacomagick 4mo ago
  LLMs are able to talk for a very long while now. Have you ever used Gemini as an assistant? It can even put in related images about your query and while it can use an improvement it can spell more or less.
  • rwmj 4mo ago
    It's far from clear that this helps you to learn anything. More likely it's a way for most people to avoid having to think or learn.
    • tacomagick 4mo ago
      I think it heavily depends on person, generalizing it as a tool that avoids thinking and learning is also wrong because I have personally used it to tacke some subjects myself and it helped me learn quite well.
      • hparadiz 4mo ago
        I'm honestly jealous of the kids. When I was 13 I was looking at books in the school library from the late 70s and early 80s about astronomy. They had beautiful shots from Voyager 1 and 2 and lots of illustrations but ultimately there was very little math in there and not too much hard science besides some basic statistics. I would have loved to have a conversation with those books.
        • KittenInABox 4mo ago
          The thing I think is underspoken in this space is that LLMs will always hallucinate a bit. How will you know as a 13 year old that an LLM is not conversing truth at you?
          • hparadiz 4mo ago
            My teachers were frequently wrong too and spoke with authority on subjects in hindsight they were frankly ill equipped to teach. Part of learning is understanding how to reason through these types of issues. It's a common problem solving problem in the work place just the same.
            • beepbooptheory 4mo ago
              Yeah but the teacher says you're wrong even if you're right. The AI tells you you're right either way. The former can facilitate an important lesson, the latter doesn't ever give you the chance to.
              • hparadiz 4mo ago
                I've been doing things like accounting where I upload receipts and have the LLM adjust a Google sheet with the money balances. The error rate over the past year has dropped from occasionally to never. That is because there's sub agents now running that check the work. If you have multiple LLMs running with a 94% success rate but you throw them into a group that requires a consensus suddenly the number basically hits 99%.
                • beepbooptheory 4mo ago
                  We simply need to run sub agents on the children's learning, then we will maximize pedagogic efficiency to 99%.
            • somenameforme 4mo ago
              I think this is a false equivalence, by some extreme margin. Teachers may occasionally have some fact or whatever incorrect, but especially at lower levels it's going to be exceptionally rare, and often based on a logical foundation that is otherwise invalid for some reason, like whether 1 is a prime number or not. By contrast LLMs constantly get things wrong and once they get something wrong will begin weaving that into everything create entirely fake realities of the sort that is more akin to a schizophrenic than somebody being mistaken on this fact or that.
        • qsera 4mo ago
          > I would have loved to have a conversation with those books. I am not sure. It would be like sending a kid to a beautiful garden with full of life and stuff, and then micromanaging what they can do there! The beauty of the books is that the books talk to you, and you cannot talk back. You have to talk to yourselves, go down some wrong path, and course correct on your own at some point, and that is where true learning happens..
        • somenameforme 4mo ago
          Hallucinations aren't going anywhere. So you're having a conversation with a stochastic parrot that occasionally says something completely wrong but completely viable and in a highly compelling fashion. The net outcome there is going to be highly negative.
          • froggit 4mo ago
            > The net outcome there is going to be highly negative. Agreed. With the way LLMs often are quite artuculate and confident sounding, it's only a matter of time until kids will be suspicious of actual people that are articulate and confident from putting effort into improving their speech and speaking about things they are right to be confident about due to being proficient in the subject. Either that or kids will see that LLMs aren't "getting punished" for being "convincing liars" and will grow up convincing the socially correct way to interact with others is by speaking to them respectfully while going all in on long-game gaslighting.
    • fatata123 4mo ago
      [dead]