alt Hacker NewsHow I use LLMs to learn complex topics
Comments
I've had success using the socratic method. I give Claude some topic (say, how the intricacies of the bond market works, the content from which are screenshots of pages from a textbook), and then I go on a walk chatting with it in voice mode. Claude is the expert, I am the student. Claude asks me questions, leading me to an answer logically. I come back with questions, and we back and forth. LLMs arent like they were in '23-'25. I'm almost always skeptical its going to lie, and almost always wrong.
In particular:
- I limit it/encourage it to give me single sentence questions
- I sometimes will ask it to tell me a motivating, human-grounded story, when we're starting a new concept: claude responds "Maya is a bond portfolio manager, and her boss has asked her to quickly price in what happened if yields go down. She knows her bond's average duration, a measure in time, but she doesn't have a percentage, which is what her manager wants. How can she give him a percentage number with just a duration figure and the proposed new yield?"
- I'll often ask claude to let me work through it, to derive the thing myself, often resulting in a string of thoughts with "yes/no" trailers, to get the LLM to reply yes or no only, and avoid derailing my train of thought. If yes, my train of thought keeps going. If no, I've got something wrong.
- I'll sometimes stop and have it craft an artifact. I typically say "build me a Brilliant.org-style interactive demo of the topic", especially when we get into the realm of looking at the actual maths of a thing (for which prose and dialog is not optimal by itself AFAICT)
- I'll do this while I'm traveling, while I'm walking, while I'm doing chores.
It's so much fun.
> What you get is a beautiful animation that is 100% accurate and free of hallucinations.
I'm not sure I follow how this is actually guaranteed? The fact-checking process mentioned just seems to involve asking AI to review its own work.
> In plan mode (using CC, or OpenCode) I ask a model to build the foundational knowledge for X topic.
Makes sense.
> I ask it to review the accuracy of the knowledge base it built in the previous step.
Ooookay that sounds good.
> I proceed asking it to build a simulation of that topic in a low-poly, Rollercoaster Tycoon-like animation.
wat.
What’s everyone’s opinion on learning new tech things in this day and age? My opinion swings between positive and depressing vision of the future.
I still learn new stuff, but I’m afraid it won’t have any value in a year or so.
For example, I’m pretty good at optimizing low level stuff, but right now you can just ask LLMs to do so and they are pretty good at it. They will profile the code and suggest reasonable options like 90% of the time.
The biggest thing I've learned from doing stuff like this is that there are no shortcuts. At some point or another, to truly learn something deeply, you've got to dig in to the boring details and do things the hard way. LLMs can help with this...but I find it's usually tempting to try and just offload the boring stuff to them, which doesn't work.
I’ve been using LLMs to create readable rewrites of RFCs and specs that interest me. It is not precise enough for implementation use, but it has increased my understanding of the underlying RFC.
Another useful approach has been asking Codex to implement complex things, like a Kademlia DHT or BitTorrent client in a literate style with the explicit purpose to increase understanding by reviewing the source code.
Examples: https://rickcarlino.com/notes/note-dump-and-ai-summaries/ind...
> What you get is a beautiful animation that is 100% accurate and free of hallucinations
How does he know?
My main issue with using AI as a learning platform is that unlike documentation, books, Youtube videos, there is not really a process of having someone "review the learning material". For example, I can always read the review of some book or ciriculum, the comments under a video, or if it's some for of open source documentation you can check the PRs and verify to some extent it's claim. With AI I can't really say what it has halucinated, because I am learning a new thing, I don't have that benefit of previously reviewed material.
The title is not representing what the post is about. “Use LLM to learn complex topics” here actually means that the author asks an agent to describe the problem area, and then implement a simple web-based simulation game, and by playing that game, the author actually learns about the topic and its constraints. They use chip making as an example.
That's actually a fun way to learn processes!
But AI will be better than you at those topics as well, and when someone needs an expert in that topic take a guess who will they approach in such scenario.
I don't think we are even that far when the complexity AI can handle surpasses 99.999% of what humans can handle, where AI make e.g. physics discoveries beyond the grasp of most humans and it will have to "dumb it down" when talking with humans -even physicists- but not with other AIs
I guess we all learn in different ways. I prefer reading, and can fairly fast find the articles that make sense to me.
Colleagues often suggest podcasts and videos - I very, very rarely listen to them or see them.
The bandwidth is too low. It's not efficient and ultimately I'm bored.
This is a nice project, it looks cute. I watched some of the pages But I want more than that, more information, and faster - still a Wiki fan.
Also, step number 2 in the flow: have the LLM check itself... Naah, I don't believe that.
But you're not the only using gen ai like that. Take care.
I’ve written a skill that I basically feed what I’m looking to do, some ideas I had for accomplishing it and any other details like tech stack, etc.
The skill then riffs with me, judging my ideas and suggesting alternatives. We go back and forth until something useful comes out of it. This process isn’t unlike how I do normal development.
However, once agreed it breaks the work into “steps”. It then creates a tutorial for me, for those steps, explaining each line, why each change happens etc. I can then ask questions, muse about an alternative idea etc. Then I do the steps, and I’ve learned and gotten what I wanted to get done.
This has been how I’ve been learning Godot and making a game for the past month or so. I didn’t go in blind, I started with a course from GDQuest so I could feel confident guiding the tutorials. I will say though, having a tutor to bounce ideas off of has been really useful.
I still try to figure it out myself, consult the docs, discord etc. But if I’m stumped I’ll run my tutor skill and have some fun.
The main bottleneck as an engineer is no longer writing or testing code. It is how long it takes to understand complex systems. This is a really nice approach that, if you have the tokens and the patience, feels like I great way to learn something and I think we'll see more and more stuff like this.
I had a similar realization a few months back and am working on a tool that generates "mermaid walkthroughs". It is 1000% less pretty but it is fast and is pretty good at explaining how services work or what a code review does or just as a way for your agent to explain some decision to you.
I use it by telling it my background, giving it a rough timeline and asking it to create a learning timeline, save progress along the way, and git push / pull periodically so I can use the same thing on both Linux and Mac. I tell it for each phase in the learning timeline, present me information, then challenge me on it. If it's code, it challenges me with a coding challenge, where I use it in an IDE plugin. If I'm learning something that isn't strictly code, then I ask it to give me info, then challenge me with questions and grill til I get it right. I ask it to save what it think I struggled with, so that later we can drill it again and I can also review it in an .md file.
Does anyone else use Claude like this?
It's sped up my learning by 10x. I struggled with 'just reading a book.' Take kubernetes. I hemmed and hawed and spent years periodically reading some dry book or blog or official doc, falling asleep, and forgetting while I got busy. Now I'm aggressively working with it, almost like I'm addicted to a gamification, of getting through our learning timeline, and I'm excited to move forward as quickly as possible and pass its tests.
It's like a fake teacher, because I can also ask it to drill into a topic or re-explain itself if it made no sense.
The only thing that worries me is, sometimes I'll say something like, "Um, are you sure about that?", and it'll apologize and correct itself. I barely challenged it!
I've been using them by reading some docs/wiki/tutorial, then when I think I understand something trying to do a rough explanation to the LLM and ask if I'm right. I'm usually making some analogy to something I already understand a little. I'm usually partially right but missing some key bits at the first pass. I go back and forward asking for explanations of various bits or asking for resources around the area I'm not understanding. Often times just discovering the relevant name for the area of study opens lots of doors. I basically use it like I would talk to a knowledgeable and patient teacher.
As for how useful it is to understand thins, I believe it's still useful and hope it will continue to be.
My high hopes were quickly dashed at the step that involved turning a complex topic into rollercoaster tycoon.
The little tool it outputted is nice, but click around the stages and the text is not high quality at all. The snippy titles, abbrievated explanations, I wish a few more iterations and thought was put into the actual main textual content. Especially for 'complex' stuff
> In plan mode (using CC, or OpenCode) I ask a model to build the foundational knowledge for X topic. I ask it to review the accuracy of the knowledge base it built in the previous step.
> What you get is a beautiful animation that is 100% accurate and free of hallucinations.
How do you make that leap?
If you're using LLMs to learn or for research, and at some point you don't end up engaging with an actual resource (books, papers, lectures, web pages, etc) then you're playing yourself.
I use LLMs to learn deep technical concepts. I really like them because I can spend countless hours a day understanding things and building an investigation file with all my findings. I code examples and test the findings. It has helped me understand basically anything.
I'm using LLMs right now to build a terminal browser, a GUI browser, and a PyTorch/LibTorch replacement. It's really fun to be able to learn and make progress this way. It's like reading multiple interactive books, where every concept can be explained again and again until I understand it.
I’ve been doing something similar (browser only interactive courses served from GitHub pages) to teach me topics from beginner to advanced
LLMs and systems intersection - https://kernelspace.naigap.com
Distributed systems - https://byzantine.play.naigap.com
Gamifying the presentation could make topics more accessible to others. For me the overhead wouldn't help with my own learning. Also I've been burned by just learning things mechanistically (e.g., coding, applying algebraic rules), so I'm leery of learning just by making flashcards or models of the topic.
I find LLM's do great for learning when I ask what are the principles, how the main applications work, what are the key drawbacks, where are the growth plates in the field, etc. - the kind of thing a good advisor points to. Sometimes I have to ask it explicitly to use topological order of topics and show relations, which often highlights the gradient changes in the learning curve. For pruning, it's surprisingly good applying philosophical heuristics - Occam's razor, or Derrida's differance (the difference that makes a difference), etc.
And finally, no learning is effective without problem sets, and for those LLM's at times get me over blocking issues.
The degenerate case is memorizing the glib phrases regurgitated back to me; they're helpful and functional enough to get me into real trouble!
I wish there was a LLM tool to explore a topic recursively, as a tree or a mindmap. You would start with some high level concept (say "cryptography") then dig further and further to more specific topics.
I think that would be a way more natural way to explore than being stuck on the classic linear output of a LLM.
I'm all for this and I'm keen to try it. I certainly don't want to take away from sharing another neat use case for learning.
But
> What you get is a beautiful animation that is 100% accurate and free of hallucinations
100% free of hallucinations when you're not an expert that can check it is impossible. LLM hallucinations are an unsolved problem.
Something that I have realized recently is that it has become so easy to get an answer to almost any question with the help of chatbots that its almost unnecessary to spend any effort thinking about the problem or the solution. I feel like before when I had to spend time researching a problem to find an answer I learned so many things around the topic itself which helped me understand the problem itself better and gained a deeper understanding. Today it feels like you can have an answer to the most complex questions you might have, yet you gain a superficial understanding of the topic and might forget about it quickly.
Personally I'm excited about these sorts of experiments. We all learn in different ways, and these sorts of techniques allow us to create "on-demand" syllabuses and lessons that fit our learning style and learning level.
It's not perfect, but I'm optimistic this will be a useful way to teach/learn in the future.
And, to be clear, I think this will be best utilized within a group/community setting. I don't think it will replace teachers or classrooms.
> I ask it to review the accuracy of the knowledge base it built in the previous step.
> [I ask it to build an interactive thing]
> I then push it to a new repo and enable GitHub Pages for it.
Congratulations. You are an echo chamber for LLMs. Use it to create, and verify, and post to then be scraped and trained on again.
This guy is severely milking it now. If learning means building an inaccurate and incomplete understanding of the topic then go hog wild. Otherwise https://news.ycombinator.com/item?id=49209049 sums up my feelings about the author's attitude.
This is a great share, gonna come back to it.
Also worth mentioning that Matt Pocock has a /teach skill that creates interactive, learning sites for learning a new skill.
Really neat idea, I think it is one of the best ways to exploit the combined building and explaining capabilities of LLMs. I am currently building an app/game to explain friends and family concepts around wealth management and wealth building. Games are a great way to hide complexity while still including it in the « guide » you are making.
I tend to ask the LLM for a single HTML page explanation, with a pedagogical approach. Something about dropping the word pedagogical leads to a more structured outcome, but I haven't quite figured it out why yet.
i looked at the animations, they look cool, and i don't think i will enjoy learning things that way. as someone else said, there's a lot of content already produced on these topics. i also think the level at which these animations are playing, they are actually hiding the 'complexity' of these topics.
The biggest win for AI dev efficiency is cutting down what gets loaded into context. Semantically matching tasks to the top tools helps a lot.
Very cool, I like the visual learning nature of this and the auto play once starting. The game graphics are engaging which counts for a lot these days, I feel my attention span suffering after using agents for the past year.
I've been working on a similar process of pushing to github pages, but focused more on having "practice sessions" with coding blocks to test content. Using webassembly and mock servers to mock backend endpoints Here's one I built to build a full stack llm chat system in the browser.
https://model-systems-labs.github.io/latent/llm-systems/less...
> What you get is a beautiful animation that is 100% accurate and free of hallucinations.
How do you know if you're learning this for the first time? Very risky to learn from LLMs. I've done it, but you have to keep your wits about you. Lots of "oh of course you're right - what I just told you was completely wrong".
i've asked llms to write a presentation for me on a given topic. for whatever reason, that finds the hidden layer responsible for explaining things well.
The selected topic (chip manufacturing) is being presented at a shallow level at best. Go read Wikipedia’s article [0] and see how deep and broad the topic is. It also skips quite a few steps, and utterly glosses over how insane of an accomplishment EUV is.
This is my biggest societal issue with LLMs: they allow you to think that you’ve “learned” a topic because you read a lot of technical terms. I don’t think it bodes well for the future.
0: https://en.wikipedia.org/wiki/Semiconductor_device_fabricati...
My favorite way to learn infra topics at work right now is asking for a humorous analogy involving monkeys and bananas. I tend to remember the result, and it gives me reference points for new topics.
LLMs can help you understand a language, but they can't replace learning the vocabulary. Words and phrases still need to be learned the old-fashioned way: repetition
Oh, I do similar, but using single page websites.
Last month, I read The Prince and had it make a text adventure campaign for me.
For a lot of other topics, I often just ask it to create a simple python example that I can run.
Building learning artefacts is under appreciated.
Inthink this was always the best way to learn. But it used to require immense work for a teacher.
My compiler course is a great example - program in plug in a stage of a compiler.
These exercises can be made on demand and incredibly easy now.
I don’t think I really agree with the author’s approach here, but I will say LLMs have been a huge help to me as I’ve been reviewing linear algebra and diving into signal processing. Anything in a textbook that I don’t fully grasp or am confused about, I just take a snapshot or copy paste then ask a model to derive it or explain it in different terms.
It reduces friction a ton, but at the end of the day I’m not skipping anything.
YouTube has so many truly wonderful videos on chip production. I admire your approach but it seems like a lot of people are in this ai maxxing phase where they reach for ai for everything despite their being ready, high quality things already available for free
Kids use agents for exam training, rather than cheating. Which is great as long as the LLM does not hallucinate
Too over engineered, few text paragraphs would suffice to understand the topic.
AKA - How I skim read a lot and learn nothing of value.
Is there a way we could turn this into a reusable skill?
That seems like a terrible way to learn. It’s a neat animation but cmon, there’s like educational TV programs from the 80s that explain this so well, in Germany there’s “Sendung mit der Maus”, not sure if they have a segment on chip manufacturing. But in these clips you can at least see the real stuff instead of some half wrong animation, just let the LLM write a few paragraphs for you or better find an ACM article or book on the subject, probably still takes less time than coming up with that animation…
God does everything have to be productized and glorified as if you’ve invented a new way of learning. Read some books!
anyone using LLM to learn a new language? got any useful prompts for this?
I thought LLMs were a great tool for learning new topics - perhaps even complex ones. But overtime, I have had several frustrations with this. First, I get exhausted reading LLM prose. I really don't want to read anything generated by something like Opus 5 at this point. Second, as I dive deeper, I need a way to organize the information in a useful way as I begin to branch out in many different directions. I have tried to use the LLM to fix this by having it generate a web page with diagrams and organized information flow. It's an improvement, but I still run into the issues I described in my first pint - LLM prose is annoyingly dense, and the useful information gets lost in a bunch of noise. You can direct it do something like "use plain English and avoid LLM prose - provide only as much information as necessary to demonstrate the point", but it is once again only a marginal improvement.
And then I begin to think to myself that I should just read a book on the topic written by a trusted source who put a lot of effort into teaching the topic properly and presenting the information in a thoughtful way. So, I am back to books and mostly try to use LLMs to clarify certain questions or ideas I have.