I am increasingly convinced that the people who claim AIs are useful for any given subject of any import (coding, art, math, teaching, etc.) should immediately be regarded as having absolutely zero knowledge in that subject, even (and especially) if they claim otherwise.
From what I can see in my interactions with LLMs, the only thing they are actually decent at are summarizing blocks of text, and even then if it’s important you should parse the summary carefully to make sure they didn’t miss important details.
I mean… Yeah? Most explanations aren’t great compared to a comprehensive understanding in your head, you already understand it - it would have to be extremely insightful to impress me at that point
The results vary greatly based on the prompt too - not only that, it changes based on the back and forth you’ve already had in the session
It’s not a god, it’s not a human expert, but it’s always available, and it’s interactive.
It doesn’t give you amazing writeups, but (at least for me) it makes things click in minutes that I might need an hour or two to understand through reading up on it. I can get a short summary with key terms, ask about key terms I don’t know, ask for an example in a given context, challenge the example for an explanations of how the example can be generalized, and every once in a while along the way I learn about a blind spot I never realized I had
It’s like talking to a librarian - it gives you the broad strokes of a topic well, which prepares you well enough that you’re ready for deeper reading to fill in the details.
It doesn’t replace a teacher, a tutor, further reading, or anything else - but it’s still a fantastic education tool that can make learning easier and faster
To be honest, I think the world would be a better place if all the money now poured into “AI” would be spent on expanding access to libraries and librarians for everyone.
so the LLM is worthless if you already understand the topic because its explanations are terrible, but if you don’t know the topic the LLM’s explanations are worthless because you don’t know when it’ll be randomly, confidently, and extremely wrong unless you luck into the right incantation
what a revolutionary technology. thank fuck we can endanger the environment and funnel money into the pockets of a bunch of rich technofascists so we can have fancy autocomplete tell us about a basic undergrad CS algorithm in very little detail and with a random chance of being utterly but imperceptibly wrong
I don’t find the explanations bad at all… But it’s extremely useful if you know nothing or not enough about a topic
FWIW, I’m a strong proponent of local AI. The big models are cool and approachable. But a model that runs on my 5 year old budget gaming PC isn’t that much less useful.
We needed the big, expensive AI to get here… But the reason I’m such an advocate is because this technology can do formerly impossible things. It can do so much good or harm - which is why we need as many people as possible to learn how to use it for what it is, not to mindlessly chase the promise of a replacement for workers.
AI is here to stay, and it’ll change everything for better or worse. Companies aren’t going to use it for better, they’re going to chase bigger profits until the world burns. They’re already ruining the web and society, with both AI and enshitification
Individuals skillfully using AI can do more than they can without it - we need every advantage we can get.
It’s not “AI or no AI”, it’s “AI everywhere or only FAANG controlled AI”
yeah, you’re still doing everything you can to dodge the articles we’ve linked and points we’ve made showing that the fucking things just make up plausible bullshit and are therefore worthless for learning, and you’ve taken up more than enough of this thread repeating basic shit we already know. off you fuck
deleted by creator
I am increasingly convinced that the people who claim AIs are useful for any given subject of any import (coding, art, math, teaching, etc.) should immediately be regarded as having absolutely zero knowledge in that subject, even (and especially) if they claim otherwise.
From what I can see in my interactions with LLMs, the only thing they are actually decent at are summarizing blocks of text, and even then if it’s important you should parse the summary carefully to make sure they didn’t miss important details.
deleted by creator
I mean… Yeah? Most explanations aren’t great compared to a comprehensive understanding in your head, you already understand it - it would have to be extremely insightful to impress me at that point
The results vary greatly based on the prompt too - not only that, it changes based on the back and forth you’ve already had in the session
It’s not a god, it’s not a human expert, but it’s always available, and it’s interactive.
It doesn’t give you amazing writeups, but (at least for me) it makes things click in minutes that I might need an hour or two to understand through reading up on it. I can get a short summary with key terms, ask about key terms I don’t know, ask for an example in a given context, challenge the example for an explanations of how the example can be generalized, and every once in a while along the way I learn about a blind spot I never realized I had
It’s like talking to a librarian - it gives you the broad strokes of a topic well, which prepares you well enough that you’re ready for deeper reading to fill in the details.
It doesn’t replace a teacher, a tutor, further reading, or anything else - but it’s still a fantastic education tool that can make learning easier and faster
To be honest, I think the world would be a better place if all the money now poured into “AI” would be spent on expanding access to libraries and librarians for everyone.
deleted by creator
so the LLM is worthless if you already understand the topic because its explanations are terrible, but if you don’t know the topic the LLM’s explanations are worthless because you don’t know when it’ll be randomly, confidently, and extremely wrong unless you luck into the right incantation
what a revolutionary technology. thank fuck we can endanger the environment and funnel money into the pockets of a bunch of rich technofascists so we can have fancy autocomplete tell us about a basic undergrad CS algorithm in very little detail and with a random chance of being utterly but imperceptibly wrong
I don’t find the explanations bad at all… But it’s extremely useful if you know nothing or not enough about a topic
FWIW, I’m a strong proponent of local AI. The big models are cool and approachable. But a model that runs on my 5 year old budget gaming PC isn’t that much less useful.
We needed the big, expensive AI to get here… But the reason I’m such an advocate is because this technology can do formerly impossible things. It can do so much good or harm - which is why we need as many people as possible to learn how to use it for what it is, not to mindlessly chase the promise of a replacement for workers.
AI is here to stay, and it’ll change everything for better or worse. Companies aren’t going to use it for better, they’re going to chase bigger profits until the world burns. They’re already ruining the web and society, with both AI and enshitification
Individuals skillfully using AI can do more than they can without it - we need every advantage we can get.
It’s not “AI or no AI”, it’s “AI everywhere or only FAANG controlled AI”
yeah, you’re still doing everything you can to dodge the articles we’ve linked and points we’ve made showing that the fucking things just make up plausible bullshit and are therefore worthless for learning, and you’ve taken up more than enough of this thread repeating basic shit we already know. off you fuck
deleted by creator
that statement being true is quite probable: it was likely impossible before this point to set this much money on fire this pointlessly!