I think learning is actually a very underrated application for LLMs; they can be extremely useful for that. However, it’s difficult to get right and requires an active effort and discipline from the learner, even in purpose-built harnesses. I don’t know if you can use LLMs to learn if you don’t already have the skills to self-teach effectively.
Additionally, I’d hesitate to let corporations shape developing minds more than they already do… Ideally, we should be reducing corporate influences on education, not integrating them deeper with direct 1:1 contact with students.
Using an LLM is exactly the opposite of learning. It’s anti-learning. It feels nice because your brain is not engaged, and it gives a lot of text that looks like explanation, so you feel like you’ve learned something, but you just skimmed some bullshit and that’s it. In the end you know less than you knew before you started, but you don’t realise that.
Not sure if I buy this argument. Couldn’t you say the same thing about using search engines to look up information on a topic? That you just “skimmed through some bullshit”? Not everything you find on the internet is necessarily true and the same applies to AI
That’s what people doing. That’s the reason they’re talking at chatbots, because chatbots mimics speech. They do “chatbot, explain X to me”, chatbot generates words that look like a text full of whatever, user skims those words, sometimes doing “I don’t understand Y part of it, explain it again but using different words”, and go “oh, I understand X now, chatbot called me a good boy in the end, this means I’m an expert on X actually”, and go about their days thinking they know shit now, when in reality they did unlearning.
I’m sure you’re different, but if so you’re unique. And maybe, just maybe, you also don’t realise how not unique you are.
It’s hard to overestimate how effective an llm is at making people believe they weren’t lied to
Perhaps if you’re building upon 15 years of knowledge and have good judgement about what to believe they could potentially be decent for self-learning, but i don’t see how that would be any easier than just doing self-teaching the old fashioned way.
If someone is starting with 0 years of knowledge, I seriously doubt an LLM would be beneficial in any way.
There are studies about the phenomenon of people not learning as much when acquiring the information becomes too frictionless (i.e. “what’s the point of knowing something when I could just ask the computer”).
I am adept at self-learning, and I would never utilize such a tool in my studies. I also don’t see any need to augment my self-teaching process, as it is already extremely easy, and I am already good at it; I don’t need assistance from software that is incapable of thought (critical or otherwise), and doesn’t “know” anything.
I think learning is actually a very underrated application for LLMs; they can be extremely useful for that. However, it’s difficult to get right and requires an active effort and discipline from the learner, even in purpose-built harnesses. I don’t know if you can use LLMs to learn if you don’t already have the skills to self-teach effectively.
Additionally, I’d hesitate to let corporations shape developing minds more than they already do… Ideally, we should be reducing corporate influences on education, not integrating them deeper with direct 1:1 contact with students.
Using an LLM is exactly the opposite of learning. It’s anti-learning. It feels nice because your brain is not engaged, and it gives a lot of text that looks like explanation, so you feel like you’ve learned something, but you just skimmed some bullshit and that’s it. In the end you know less than you knew before you started, but you don’t realise that.
Not sure if I buy this argument. Couldn’t you say the same thing about using search engines to look up information on a topic? That you just “skimmed through some bullshit”? Not everything you find on the internet is necessarily true and the same applies to AI
That’s not learning, that’s just reading LLM generated prose. Not entirely pointless depending on the circumstance, but not what I’m talking about.
That’s what people doing. That’s the reason they’re talking at chatbots, because chatbots mimics speech. They do “chatbot, explain X to me”, chatbot generates words that look like a text full of whatever, user skims those words, sometimes doing “I don’t understand Y part of it, explain it again but using different words”, and go “oh, I understand X now, chatbot called me a good boy in the end, this means I’m an expert on X actually”, and go about their days thinking they know shit now, when in reality they did unlearning.
I’m sure you’re different, but if so you’re unique. And maybe, just maybe, you also don’t realise how not unique you are.
It’s hard to overestimate how effective an llm is at making people believe they weren’t lied to
Learning with AI
Noooo not like that lol +1
That’s just investorbait grift. I’m a self-taught programmer (~15yrs), and I both teach and learn using LLMs frequently now.
Let me say it another way: LLMs are a useful tool for self-taught learning. They are NOT a replacement for a teacher or tutor.
Perhaps if you’re building upon 15 years of knowledge and have good judgement about what to believe they could potentially be decent for self-learning, but i don’t see how that would be any easier than just doing self-teaching the old fashioned way.
If someone is starting with 0 years of knowledge, I seriously doubt an LLM would be beneficial in any way.
There are studies about the phenomenon of people not learning as much when acquiring the information becomes too frictionless (i.e. “what’s the point of knowing something when I could just ask the computer”).
I am adept at self-learning, and I would never utilize such a tool in my studies. I also don’t see any need to augment my self-teaching process, as it is already extremely easy, and I am already good at it; I don’t need assistance from software that is incapable of thought (critical or otherwise), and doesn’t “know” anything.