Using LLMs to get advanced in a subject faster | Hacker News Reader

Using LLMs to get advanced in a subject faster

2 pointsjamesdc39 comments

I've recently have become enthralled with using LLMs to help me learn faster. As a competitive person, I'm always looking for ways to improve my knowledge and skills.

We've all been there - starting something new, feeling overwhelmed, and struggling to retain information. Traditional learning methods can be slow and inefficient, leading to a frustrating cycle of memorization and forgetting. That’s how I felt in school I would forget most of what I learned regardless of my grade.

LLMs are trained on vast amounts of web data. It can enhance our learning and help us retain information longer by teaching us in our preferable method of learning. By leveraging these models, I believe we can overcome the limitations of traditional learning and achieve our learning goals more efficiently.

Personally, I want to learn more about Poker, Crypto, Mathematics

I'd love to hear from you - do you struggle with retaining information or feeling like you're not learning efficiently? What subjects or skills do you want to improve or master?

Claude Sonnet 3.5 at least works awesomely, you can just talk to it asking stuff, it will infer your knowledge level from your questions and start answering according to it, proposing a follow up path for further insigths about the subject. If you follow it up, usually it will take you to a somehow similar path as you would follow if you just happen to having been reading the wikipedia page for the subject you're asking Claude to explain about.

but if the subject isn't something that obvious as something you can find in the wikipedia, you'll be good too, Claude will take to a sort of "shortest path algorithm" of knowledge about the subject you're looking about.

if you side it with web searching, you'll see it takes some few keywords, concepts about the subject, and explains them, and you can go deeper searching them and looking other sources (blog posts, answers in reddit, etc.).

In Claude I found almost no hallucinations in the deepest explanations it answered in some research I've done with it, maybe some non-human focus on trivial details while not looking at more relevant stuff about the subject (I infer that the training data could be cramped with the less deep data, more people answering about the subject on internet from a shallow level of knowledge than a few experts answering really good explanations > you'll find these answers first when you begin web searching).

LLMs will teach you absolute nonsense, and if you're truly a novice you'll never know until it's too late.

If you want to actually learn things (eg. pick up lasting marketable skills), you should read from published professionals and get official credentials. Learning requires putting in the work, shortcuts will compound as you go further along and leave you more confused than your peers. Don't go around mistaking yourself for the self-taught professional if you lean on AI as a crutch.

Must upvote this. I sometimes ask Copilot instead of startpage.com especially when I find it hard to formulate those kind of keywords for a classical googly search engine that would probably lead to meaningful results, and LLMs can be very helpful with that. The other day I had copilot write me a bookmarklet that will cause the focused DOM element on a web page get highlighted with an outline; nothing compared to those success stories you get to read but I find it impressive it did just work—copy, paste, done.

But the downside is that when you ask questions about a subject matter that you deeply care for and are well-versed in, or when you have specific formal requests (like the other day when I asked for a semicolon-delimited enumeration of terms and it kept giving me lists with no semicolons), or sometimes when you ask the LLM to avoid certain terms or stylistic mannerisms—those are often moments when the cracks in the system become glaringly obvious.

And that's where the problem lies when using LLMs to learn new things, you never know where it goes off the rails. Sure, that random blog post on the internet has much the same problem; maybe it's just because of extended exposition that I tend to believe I can rather tell an incompetent blog post from a relevant one than doing the same with answers by ChatXYZ.

So.. "Brodies Notes" or "Lambs tails from Shakspear"

if you want digested knowledge, I'm not sure an LLM is the best path into a digestion of knowledge but I am also unsure digested knowledge is deep learning. So your competitive "know more facts" approach sounds to me like classic autodidact model: Learn facts, assume your inferences over those facts is "knowledge"

I cannot count (there are too many) the number of times my own autodidact instinct was .. completely wrong.

I'd not worry too much about forgetting things when learning; it's part of the process. Immersion, repetition, and, above all, a vivid, maybe visceral interest in the subject matter are essential.

The best way to use an LLM to learn a new thing:

* figure out what you'll use to check to see if the LLM is correct

* go study THAT instead

i learned all about keto from LLMs. It was quite informative.

what kind of rocks are keto?

i don't know what you mean dog.

Reply on news.ycombinator.com