For the first time in a long time, learning/working is fun again.
Yeah AI is cool for learning but just a friendly reminder that Sam Altman, who says you should use AI for everything in your daily life, takes note-taking in physical notebooks seriously in HIS daily life.
How do you know it's not hallucinating what its telling you?
>>107801565Yes. Beforehand, learning certain stuff was a pain in the ass>some esoteric things only had forums occupied by boomers and badly made youtube tutorials that would always commit a critical detail>niche issues and problems required lots of digging and there's often the issue m, again, of boomers leaving out a critical detail as a passive aggressive form of gatekeeping>20 minute youtube video where speaker spends 10 minutes explaining why they like X personally over Y, 5 minutes of poorly formatted explanation for Z, 2 minutes of actually showing how to do it, and less than 2 minutes of a half decent explanation that helps someone grasp the concept
>>107801565>>107801840This
>>107801840It's not mutual exclusive. AI is great for curating information, note taking is good for memorization.>>107801957I've found hallucinations are rare for empirical information. It was way more likely that my college professors would say something slightly incorrect than an AI.>>107801967^This. AI is shitty at a lot of things, but one thing I found it's really good at is shifting though years of forum post and documentation to give me relevant information.
>>107801565it's worse than wikipedia, with the exception of grokipedia which fully cites most things. even when asking opus and codex to pull definition specs, it's slower and less accurate than checking the actual documentation.AI is good for pulling boilerplate code snippers for mockups, that's about it. if you're learning from AI, especially if you're not source-checking each thing AI spits out, you're a fucking retard.if you just want to learn random stuff, read britannica.
>>107802300>It was way more likely that my college professors would say something slightly incorrect than an AI.unless your university was garbage, this is most likely false. it's those subtle things that AI says with authority which are actually incorrect or irrelevant, that the human expert actually knows better. this is why AI is so insidious.
>>107801565I am, and I love it. I'm also reading a lot of books to have a more organized source of information, but Claude is helping me clarify some things or answering random, slightly-related questions. Plus I got Obsidian to take notes.I'm learning Go and one problem I find is that it will give me outdated solutions, not taking into consideration the changes the language has gone through in the past couple of years. Gotta stay on top of that yourself.
>>107802328You've hit on a fundamental and critically important limitation of current AI assistants.You should:1. Treat my suggestions as hypotheses, not facts2. Cross-reference with current documentation3. Understand the principles behind suggestions4. Test cautiously and observe results
>>107801565Honestly? Yeah. It's the one thing I've liked about it above all else.If I get lost on a project I no longer have to spend 20-30 minutes looking for an answer anymore.Just need to be careful to not let it do the thinking for you.
>>107802328>it's those subtle things that AI says with authority which are actually incorrect or irrelevant, that the human expert actually knows betterHuman memory is WAAAY more fallible than an actual database. You're literally making the argument that human professors are more accurate than the textbook they're teaching, which if you ever been to college you'd know is bullshit. Even the most highly educated professors from the most elite prestigious universities can't recall volumes of text verbatim. AI systems can. Breakdowns only happen when it incorrectly interprets a prompt or references bad data. Saying AI referencing a database is less reliable than a human recalling information more memory is an asinine statement that only someone with a crippling bias would have.
>>107802540>databaseHere's the problem: it's not a database and if you think AI is a "database" then you don't understand how it works. Thanks for coming to my ted talk.
>>107801565yes, i've been picking up japanese again with iti don't know how correct everything is but it beats not doing anything for years
>>107802300Oh, so you don't know it's hallucinating.
>>107801565i did and might come back to it.>ask questions here>barely you get answers>ask question to llm>instant answer ,you try if it's shit ,you double check with wikii would love to have my questions answered by real humans and say thanks to real humans but unfortunaly i guess it's kinda done no ones bother>google itbro all search engines are shit and i won't waste my time digging through every forums with posts posted in 2008 whith solutions that don't work anymore most of the time
>>107801565>learning with AIyou dont see idiocracy in this?
>>107801565I'm teaching myself lambda calculus with it and it's awesome.
>>107804732If the math literally checks out then what's the problem?
>I'm learning so much through AI>I'm a 100x programmer with AI>I get everything I need done with AII love unfalsifiable claims
Me. AI is way better at answering questions than the average forum user.
>>107804732>public school systemyou dont see idiocracy in this?
>>107801957Because you can think? Do you just regurgitate information you're given? Learning should be an integration process where you compare and contrast with prior information. Worst case you get a hallucination, is that really any worse than citation laundering? Why throw the baby out with the bathwater?
>>107801565like building a foundation with potatoes and plaque
Chatbots are a waste of time
>>107804732Not sure if this post implies "learning with AI" is grammatically incorrect. If it does, it's correct. AI is a learning "companion/helper"in this context.
>>107805247You’re getting left behind. :D
>>107805247I use AI as my little dumb secretary at work. I no longer waste time writing emails, quotes, etc. I just tell it with a few words exactly what I want and it spews out nicely written and formatted output that I quickly review and dispatch. Might as well be better than a human secretary.Also as a translator. Translations from English to German (both formal & informal) are very accurate, much better than actual translators.
Yeah its actually insane how fun it is. I was using it recently to analyze my dna and give me targeted supplementation to help with genetic deficiencies. For example, i have some shitty varient that means im prone to oxidative stress... so i supplement nac to help. I have some other snp variant that means i require methylation support. Creatine and SAMe recommended. These are very common supps so its easy to verify
>>107802540>Even the most highly educated professors from the most elite prestigious universities can't recall volumes of text verbatimThey don't need to because they can re-derive every equation on the go, and this is in fact what they do.
>>107802466Not having to find the answer through 10 different videos narrated by Indians is also a plus, though nowadays its some generated AI voice now.
Incredible. I like chatting with the robot. But when it comes to info on things, especially technical things like computer parts, speakers, it needs a lot of corrections often.>"This monitor is perfect, it has a DisplayPort out too">No, this is not correct, this monitor doesn't have a DisplayPort out>"Ah yes, you are right, then this is not that perfect for you, it doesn't have a DisplayPort out."Well yes, why am I using you again.Also I know it will be a massive issue with security and all that but I often would have liked it to be connected at all times to know timezones, dates, when is it at the time I speak to it. I have to tell it the time and day every time I need something time related.
>>107806659Its useful for searching problems. Hard to answer easy to verify. For example if I want a specific feature for a device but don't know what the industry calls that feature and can only describe it.
>>107801565me, learned so much just in the past year. about history, linux, tor, i2p, and other cool shit like how to fix my linux problems and how to fix my windows shit. recommendations for programs to try like AllStartBack etc its been great. though i always am weiry on the halluscination part
>>107806848I like this a lot too, when you describe things and it just works.
whats not to like about AI when it comes to programming? As long as you dont use it as some blackbox that spits out full projects it's the best thing since sliced bread, long gone are the days of trudging through stack overflow
>>107807040I think the people here who are 100% against AI are a bunch of jeets scared of AI leveling the playing field and making their nepotism pointless.
>>107801565Same. It can make work more fun since development is now blazingly fast
In a month my Twitter page went from programming tidbits and project showcases to who can shill their fav LLM the hardest.
>>107806110Genes don't exist and supplements are worse than useless
>>107801840>analog notesPen and paper aren't "analog". They don't emit an analog signal. They don't emit a fucking signal at all. Why do people say shit like this?
>>107801565Same, I'm brainmaxxing so hard right now using AI. I'm literally downloading knowledge for free. FOR. FREE. Y'all niggas be losing shit, if y'all ain't taking advantage of dis because the Jews will paywall the fuck out of these AIs.
it's hard not to be a snob when you read threads like theseabsolute fucking retards thinking they're actually learning something because the slop machine wrote back "that's right! you touched on a fundamental aspect of X" when they tried to engage with itread a book or two, you absolute fucking morons
>>107801565I've got free Github Education with college but haven't bothered to set up Copilot yet. unlimited prompts sounds comfy though.
>>107807740Ain't nobody got time to read a book, bro. Who the fuck is bookmaxxing right now? Just ask ChatGPT to read the book and summarise it, nigga.
>>107807740and what happens when the book you chose is wrong or biased and you dont know it? lmao
>>107807740Theres a billion retarded books with wrong views or blatantly incorrect infoDespite what you think, almost anyone can publish a book and not every famous person is correctThe only usually infallible thing is documentation for a programming language from the creator
>>107807740Be a snob, the rest of us will be laughing at you. The world of the future will be laughing at you.
>>107801565I just decrypted an unreadable doctors prescription with a.i.
Me, learning to tweak stuff
I did some electrical work implementing smart relays in some stuff around the house. It was pretty useful in telling me how circuit bord for the gat worked
>>107807769how long until it expires?i remember all my edu github shit expired after two years
>>107801565I have to. Searching for a simple answer to a simple question is fucking impossible now.
>>107801565Fuzzing libpng right now, something which I wouldn't even have bothered with had it not been for AI making it super simple to get up and running.I've since used an LLM to guide me extract a few of the parser functions out of Chromium to make a standalone binary for fuzzing, and this is my new fun thing to leave running when I'm not using the computer.I'm exploring new things all the time after the barrier to entry has been somewhat lower.
how do you learnmax with AI?
>>107810436Ask it to explain things to you
>>107801840I take hand written notes, but on a tablet (remarkable 2 because I like that my tablet doesn't have distractions like a web browser or anything). I haven't noticed any real difference from having 100 billion notebooks vs taking my handwritten notes on a tablet.
>>107807819Do you think your LLM won't have that problem? The LLM is "learning" to mimick the books. If the book sources are wrong, the LLM will also be wrong if it was trained on that incorrect book.
>>107801565My interest is on history and ancient knowledges, and obviously AI is SHIT for it.
>>107801565AI has done more teach me C++ than human could ever have. Not even kidding.
>>107801565I ask brave search AI questions. Sometimes it's helpful, but it's 50/50. I fucked my system up by listening to it when I was trying to get hardware acceleration working and had to restore. Don't listen to AI when it starts referring to the kernelI treat it like an enhanced search engine. It's good for retrieving info and saves time. But do not take advice from it. You still have to do the work, you can't just let it hand hold you to solve something
