(TOK dev here.) Thanks for the linking. I'm sorry I could not keep it glorious (that is free and login-free). I got spammed by bots and I don't have time to discriminate so I made it BYOK and kept it at that. My idea deserves a better execution than this and so I don't mind if other people go at it. All the best.
BTW, related to learning AI software, I'm working on something else... Calling it "InfoRush AI" for now, it looks like that: https://imgur.com/0tncarJ
This is Python, BYOK, FOSS. Multicolumn synchronous browsing... orchestrated by the tree of knowledge. Going beyond and beyond. I'll post it on HN soon, (when it's done).
This is a hierarchical representation of any given piece of knowledge.
It starts with a tree root node that you specify (let's say Kung fu), then it branches out into multiples subcategories (techniques, styles, philosophy, weaponry, ect...) and then you can click on these subcategories to branch out even more into the graphical tree.
This is all generated on-the-fly with Claude 3.5. There is no limit to what knowledge you might explore.
The killer feature is that it is totally free and does not require to login.
Just click the link and have fun. I'll keep it free like that, as long as I can.
I hope you like it guys because it is the best project that I have up my sleeve.
As a marketing spin for this - consider packaging it into Nvidia NIM format and make it generate 3D graph as 3D OpenUSD scene. From where I'm standing this route has a lot of potential.
Also if you never looked into it there is a project called wikidata. There each object contains a unique ID and hierarchy, which helps build semantic web. Exploring their data using your interface might be effective. (please check similar projects though as an idea seems straightforward and someone might have already done that for them)
You're a legend mate I know nothing of what you say, but it is very interesting. It is time I brush up my skills and I thank you very much for the clues! I'll look it up.
Making a note to try this when it's not slashdotted (hackernewsed?).
This sort of thing is _very_ interesting to me, and I rather desperately want something like to this tied to listings of books.
I would like to read more diversely --- at one time I was trying to read one book from each LoC division, starting at the top/broadest (so A--Z), then iterating down, but this got to be exhausting because it was difficult to determine which book to read.
A later effort was to read biographies of famous people to my children in chronological order as they were growing up --- dry run was the very simple set of U.S. Presidents --- but again researching and ordering was a big problem.
So, my hopes for this are that it includes footnotes/references/links to supporting material (perhaps affiliate links on Amazon would be one way to fund things?) and that it is possible to click through to find information on specific technical topics as well as general explanations.
Thanks. I had tried that but the AI keeps inventing false books or providing fake URLs. When the AI will be proper I will include that feature, for sure!
A message to everyone: Thank you for your interest and comments and mails and votes and feedbacks. I am sorry the service was interrupted & down & I had to makeshift with ugly donation links in a ugly sidebar for hours, ect (I got 5€ donation! not crazy but it was actually useful)...
Anyway I managed to get it back running, and running strong! Mail me if you have nice words to say or if you think we can do partnerships. I received nice proposals already, which should improve the service significantly in the coming days. The goal is to entertain and educate, and as free as we can! haha, modestly. Thank you again for everything!
This seems really cool and I'm excited to play around with it once it's up and running properly again. These type of things are my favorite applications of LLMs.
A while back I made something similar in the form of an incremental "clicker" game where you split things ad infinitum: https://lantto.github.io/hypersplit/
It is back and running, and running strong! Enjoy. Your game is nifty maybe it'll give me ideas. There could be an option to hold Ctrl to split branches vertically without branching down, ect... cool
Nice work. An AI populated mind map. I like it.
Just missed a way to go from a node to more textual content if I want to know more than just a bullet list. E.g. an abstract-style paragraph, or even just a web search for the string.
Good idea mate I'll probably include a Amazon "search list" for books, and a video "search list" for Youtube, ect... Might be cool. If I try to provide direct links I get a lot of false results but I didn't think of adding web searches themselves.
(There was a feature to generate dissertations on whatever topic but I had to turn it off because I believe it was abused and that cost a lot of credits. I'll try to add it more securely very soon.)
Love this! Such an interesting UI choice. Would almost prefer new branches to appear further out to the right instead of overlapping existing words, and then auto-scroll me to the new spot to reduce clutter when browsing several tiers in.
The context sidebar is great! Maybe the first good "AI search engine" UX I've seen (most attempts copy Google look/feel which is more suited for web page results).
This is a very fun idea and I’ve had similar thoughts, without yet turning my mind to execution. Bravo. Is there somewhere I can follow more of your work in the future?
Hi cyrillite. Send me an email with "subbing" as object: pierre.treeofknowledge@gmail.com
Then I will gladly send you what I got, whenever I get it.
I will not forget. Thank you, friend!
I'll probably make a full blog soon for added convenience. (it'll be about AI products & production)
If you want a bite of some of my other work here is a AI game I made & which got totally ignored:
https://lywald.itch.io/the-fall-of-mankind
I tried to monetize it, that was a bit lame, but in the next few days I'll release it as free.
Well, since you said Kung fu, I started at martial arts and 5 clicks in it's telling me about the self defense applications of swinging from chandeliers. This is yet another pile of LLM word vomit.
> One innovative approach involves using ceiling fixtures to gain leverage and mobility. This concept can be particularly beneficial in scenarios where traditional movement options are restricted.
...
> Utilizing ceiling fixtures for movement in self-defense is an innovative method that capitalizes on environmental elements to enhance escape techniques. By integrating body mechanics with strategic use of surroundings, individuals can increase their chances of successfully evading threats in confined spaces. Always prioritize safety and practice these techniques to ensure readiness in critical situations.
I like this project but I don't think I agree on the 3 branches that I see when I open the website.
Science is not a standalone category. In fact science is directly connected to mathematics which is directly connected to philosophy (through logic for example)
We might argue that also art is an extension of philosophy (e.g. finding objective or subjective beauty).
Probably Claude doesn't agree with this definition.
I agree with you. Truth is this was the first branching generation which I generated, and so I kept it for the sake of history. If you manage to create a better "starting branching tree" which looks nice and feels nice then I might take you up on this.
I'm sorry guys but my API key is out of funds until tomorrow's or maybe tuesday's paycheck. I'm only a student and so my money is very limited. I'm sorry this is pathetic. I didn't expect to have so much trafic these last few days, and then with this thread. Come back soon!
What do you think that I implement BYOK (bring your own key), does that interest anyone so that it is never down for them but they have to pay Claude on the side? Is anyone interested?
Thanks, fastneutron! I might give you the code if you send me an email and you explain your intentions. No problem. For now, it is not full blown open source because I don't want the added headache of managing that (double checking security & quality... I have no time for this is a sideproject).
Instead of adapting to Anki, you could add your own JS script to this page, which picks a random sample from the array of tiles, highlight it in white, and you need to press the corresponding shortcut to go to the next sample (using a dictionary). You could memorize the time-to-press to generate a finale score (if you're motivated to do so).
I had implemented that algorithm to memorize videogames shortcuts. This could be a nice challenge for your skills! Maybe you could make another page for "VS Code" and that'd be niiiice.
Thanks, and thanks for sharing, I love the idea and will think about it. Even though my intuition says it would be more productive if I put more time into adding content and polish to the existing concept, and people openend the actual software to practise their muscle memory.
I won't make this for other software, because I don't have that level of expertise for anything else. I've spent years helping other blender users, this is why I feel confident to author the information in this particular case.
I googled your name and 'blender' but didn't find any training resources online. I'm a new Blender user and am on the hunt for resources beyond BlenderGuru's Donut tutorial (which is great, btw).
This is awesome - I was a Master at Autocad for decades, and it was because of mastering keyboard shortcuts for every thing I did and having a vast ustom keyboard mapping. (When I was learning Autocad in the early 90s I read one book "The Abc's of Autocad" -- and I did every lesson in that book ~five times, which taught me all the keymappings and I never had to read another book on Autocad.)
To augment your learnings for blender here are my top:
Also, AI is fantastic at creating python snippets to do things, another HNer was asking about mapping the mouse for macOS and you can remap everything in blender using python (as its all python) --
And I have the bot reference where in the blender docs it got the python functions to map things: https://i.imgur.com/l1S8JaC.png
Its a pseudonym but I dont have any educational content in my real name either, this is my first attempt to contribute.
But the amount of quality content on youtube has been increasing steadily over the years, If I had to recommend 3, I'd say Grant Abbitt (Beginner), Erindale (Geonodes), Ian Hubert (for Inspiration).
Send me an email and in a few days/weeks I might make it open source, I am considering it. In the meantime maybe you'll get a peek. It's nothing interesting, though. Fast to replicate for an intermediate programmer.
1) You can use "the reproduction of snails", or "Call of Duty mastery", or "Learning Chess > Chess psychology > Visualization techniques"... There is no end to what you can learn. The sky is the limit!
2) Thanks I'll fix that at the next update, if you mean the grey colour of the box.
3) This is a revolutionary tool because you can see how a domain is organized, so you can catch in one glance what you can learn, and select the most interesting to you, with very high Signal to Noise ratio. This is probably better than Wikipedia when it comes to strolling around the fields of knowledge. This is a bit preposterous to say, but I say it.
To add it in a academic workflow, you just type the name of your related domain, and don't be afraid to be very specific. TOK (tree of knowledge) will give you most pertinent information to help you understand and grok all of the complex information for you. I'll add a feature soon to generate quizz/coursework for any given node. Stay tuned.
I'm happy I got only 4 upvotes because I couldn't handle a heavy load for now. And I'm thankful for these upvotes. The website will stay up for the time being. If one day it is down, send me a email and I'll give you the code.
I don't see how this would help. Such protocols may not even be written down, but rather implicitly passed from mentors to mentees in security agencies. I am all for government transparency, but no amount of transparency will reveal that a cluster in Utah is in direct link with a cluster in St. Petersburg is in direct link with a cluster in Kiyv to provide unmasking services to their administrators.
These administrators can then launder the information to their respective agencies by means of any number of play-pretend activities you can write up for the transparency committee. The agency doesn't even need to (officially) know.
They say "you need to fail to succeed". The key thing is that it is not enough to fail, you also have to reflect on why you failed, and this is the reflection which has you progress.
It is not the failure itself which is progress, it is the act of failing, then feeling BAD, and then thinking about how not to feel BAD anymore. Failure without feedback is pointless failure. (This is the same idea than deliberate practice.)
BTW, related to learning AI software, I'm working on something else... Calling it "InfoRush AI" for now, it looks like that: https://imgur.com/0tncarJ This is Python, BYOK, FOSS. Multicolumn synchronous browsing... orchestrated by the tree of knowledge. Going beyond and beyond. I'll post it on HN soon, (when it's done).