Reading manuals is genuinely an mode of activity these days reserved for those who enjoy wasting their time more than doing things. I used to hate chatbots with a passion until I realized they just spit out whatever information you need pretty much instantly. Kind of a bummer that OpenAI is bleeding money so much that there is no way in hell it will be free to use forever but while it's there you might as well use it and get a head-start on whatever you're doing.
>>107112229yeah ive been VC money burning
it's a jewish butthole every time just like the arch logo is fat man
>I can't read so I am using chat bot to read for me. Manuals are too complicated thanks for the blog update anon
>>107112229>>107112266vc dont surf
>>107112315I love the smell of datacenters burning in the morning.
>>107112451*march of the valkyries echoing in the background*
>>107112283reading the output of ChatGPT is still reading.
>>107112526i have a handful of gmails signed in on firefox containers so whenever i hit a limit i just switch to the container with the other authenticated account to continue free outputs
>>107112585noice.also good to know they dont limit the thing per ip, but per gmail accountngl, this shits really usefulyou go "look up x for me" then "dress a comparison table"boom. an hour of work in two lines of promptits a powerful force multiplier
>>107112585>>107112664contsame with manuals and programmingbut sometimes the training has just wrong data like i saw with intrinsicsand the chatbot, predictably, failed where the documentation was imprecise.
it failed silently thoughit just hallucinated in the blanks so its a thing to look out for
>>107112664its a good idea to think about it too for compartmentalization. one container profile - one gmail account - one broad focus on knowledge i wanted to do that with youtube too, so you can make the algorithm work in your favor in a sense. because each profile would only be focused on that subject
>>107112764neat idea.this way you dont have recommendations from multiple subjects compete for the same spaceor dilute the signal that the algo works withi browse in private cuz im still paranoid about google banning accounts that use adblock and i didnt make any altsi would need to find a bunch of phone numbers.its not an impossibility, but its work. which im too lazy to do wo a valid reason, kek
>>107113111yea, better take advantage of it before the whole thing comes crashing down
>let's input all manuals into a massive machine that heuristically computes and outputs a new manual that may or not be right so that I can read this new manual instead of the original manualgreat idea man
With new software I look at the actual manual to see what all the software can do. An LLM will help you get something running, but it won't introduce you to the full feature set.
I'd rather spend 5-10 minutes clicking and scrolling through an 80k word manual looking for the one 2k word section that I then have to skim through and read for a few minutes until hopefully I find the one or two sentences that answer my question. It's just much better than asking chatgpt and getting that answer immediately and spending 30 seconds confirming that it's right.
>>107112279
>>107112229>I used to hate chatbots with a passionYeah, the standard low IQ /g/ay contrarian.
>>107112229>Reading manuals is genuinely an mode of activity these days reservedGood morning, Sar!
>>107115927it's insane that people actually think like this lol
>>107118898its not, really. i prefer to use an LLM myself but its entirely dependent on if that LLM can properly parse that specific material youre after. and then you end up reading it anyways.if you were running a local LLM model, you'd be waiting about as long as it took you to find the source and read it anyway.>it's insane that people actually think like this lolno it's interesting, because he's actually still thinking and we're not, because when the LLM can find the proper thing and can read it for us, it's using the context we've provided so basically we dont have to think at all, it will find the source, and apply it to our problem. whereas he has to read the source, then think and apply it to his problem.he's superior as a human, we're superior as calculators
>I used to hate chatbots with a passion until I realized they just spit out whatever information you need pretty much instantly.Only if your usecase is extremely simple. Case in point: I was writing a tool for work this week using the python framework textual and Gemini Pro insisted I must import the RichText class from textual.text where, of course, it doesn't exist. No matter the amount of prompting it would not tell me where to import the class from correctly. Now I actually KNOW my python libraries and knew it really meant import the Text class from the Rich library and so the problem was an easy fix but an unexperienced programmer these chatbots are going to mislead so brutally that its actually funny. Basically every time you even get slightly specific they start spewing bullshit and if you don't know what to actually do you are fucked with them. If you use them to learn they will subtly poison your knowledge with bullshit sneaked in here and there.So yeah I'll be sticking to docs mostly for now. Enjoy your ultimate midwit tool that will hurt you in the long run.
>>107112229>an mode
