THE. AI. YAP.

why i personally... (and overall) think ai sucks, my experience with it etc.

OK NOW THAT I FINISHED WRITING IT... i swear the more you read the more it'll make sense I SWEAR--

disclaimer: i only used the free versions, so i dunno how its like for the paid ones

lets start with...

PART 1: WHY AI SUCKS (in general)

#1. PREFRENCE.

BREATH IN

YOU. DONT. NEED. A. COMPLICATED. REASON. TO. HATE. AI!

its kinda like food! you like some food and hate others for no reason, other than prefrence. SO. PEOPLE. SHOULD. NOT. SHOVE. WHAT. WE. DONT. LIKE. DOWN. OUR. THROATS! do restaurants FORCE you to eat sht they want you to? no!

if ai slop was food, it'd be BORSHT (ew)
(thats a joke... no offence to borsht enjoyers, id say ai would be more closer to prison food)

#2. PIXEL ART

the whole POINT of pixel art is that there are tiny squares that make up a picture. where as ai pixel art is blurry and inconsistent (no squares), you can hardly call it a pixel art... heck that's generous actually you CANT call it pixel art AT ALL, its just a regular ai generated image without the "pixel" part

#3. CONTROL

when you code YOU get to name variables, YOU get to add comments as you like and YOU get total control of what you want to get! and not just coding! drawing and writing too! YOU'RE IN CHARGE.

its like cooking, sure occasional mcdonalds or pre cooked meals but isnt it rewarding eating WHAT YOU MADE?? ok this one's more subjective but you get the point...

#4. MAPS

i mean... duh. ai cannot make an accurate map. germany might suddenly get its 1918 borders (and more??), why is caspian so small now, is it messing it up with aral?? or even small ones like cities being far apart from each other or in wrong places. and... THESE ARE MAPS FOR GOD'S SAKE-- THEY HAVE TO BE ATLEAST SOMEWHAT ACCURATE!! especially when youre using it in a presentation or wanting to gauge the world

especially when you consider how EASY IT IS TO MAKE A MAP. you find a template on internet, then colour, change borders however you want etc.

#5. TECH GUIDES

this one's gonna be lengthier--

nothing and i mean NOTHING will EVER beat a human, recording steps for a solution/guide to solve a problem on an os or app. ai will mess up macOS with windows, recommend solutions for a completely different app and even if you tell what youre using and what version, it will still mess stuff up and give generalised advice!

for example ask it where to change keyboard shortcuts in libre sprite. it will halluciniate like 10 different answers while in docs or just googling, will show you a correct soltuion right away. this gets worse with niche apps

and you know whats even more sad? ai generated hallucinations are replacing actual human tutorials!!

--------------------

LIKE FOR EXAMPLE:

solveyourtech.com/how-to-get-rid-of-rounded-corners-in-windows-11-a-simple-guide/
ai hallucinated 5 SOLUTIONS. WHICH NONE OF THEM WORK T-T

now lets check a human made solution:

thewindowsclub.com/how-to-disable-rounded-corners-in-windows-11
SIMPLE. TO POINT. gets it done. admits there isnt a native solution, while ai kept hallucinating without even locking in an answer

--------------------

OR MY PERSONAL FAVORITE(s):

spellapp.com/resources/how-to-see-when-a-google-doc-was-created
and
clrn.org/how-to-see-when-a-google-doc-was-created/
BRO IS FLAPPING AROUND. goes THE LONGEST ROADS, WHEN THE SOLUTION, IS LITERALLY file -> information AND TELLS YOU WHEN THE DOC WAS CREATED!! WHY THE F*CK DO YOU NEED TO ACCESS GOOGLE DRIVE, WRITE CODE, SEARCH, ACCESS HISTORY AND WHAT NOT WHEN ITS RIGHT THEREEEEEEEEEEE

and don't tell me "well the spell one says it has a human author!" yea access this dude's profile. it literally says he helps people write documents with ai... yeah... ANYWAY--

--------------------

IVE GOT MORE EXAMPLES. like a video footage this time

youtube.com/watch?v=UypsyMvrMac
at 8:46 copilot asks foci to access a setting that doesnt even exist TwT AND COPILOT, is developed by microsoft, SO MICROSOFT'S OWN AI, DOESNT EVEN KNOW THE OPERATING SYSTEM, IT IS INSTALLED IN--

--------------------

AND NOW MY OWN EXPERIENCE: i asked chatgpt how to delete my openai account (since somehow clicking on delete just logged me out?) and bro just recommended broad locations and general advice rather than giving concrete steps. LIKE DUDE, ITS YOUR OWN PLATFORM WHAT THE FUCK--

but enough of that, lets look at some human made documentation: github-wiki-see.page/m/LibreSprite/libresprite.github.io/wiki/Know-your-workspace

beautiful, screenshotted, short documentation of the basics of libresprite. now ask an ai for that, just a wall of text (no screenshot or accurate visuals) where half of what it said doesnt even work. you have to agree these kinds of human made guides are objectively better

you might think im nitpicking here, but if you like tinkering with tech like me, the more you do it, the more you'll see these ai articles, giving completely false solutions where it'll start gettin on your nerves

the only exception is linux, since the command prompt stays farily consistent across all versions* and is VERY well documented, so ai is able to help there pretty well tbh

IN SUMMERY ai has never used an app, an os or anything for that matter and is purely going off of assumptions. human made articles documenting each step in a specific tech situation is always better

PHEW that was lengthy! but before we move onto the next part some of you are prob yelling:
WELLL ITLL GET BETTAHHH.
and there are good counter arguments-- BUT I PERSONALLY... wont... say... anything... fine, maybe it will get better. but the point still stands: currently, its not perfect! so stop acting like it is

PART. 2: PERSONAL. EXPERIENCE.

ok i talked about it on the hyperrant page but its buried deep and i want more people to know my stance n experience so eh lets put it here too

AHEM. it all started back when chatgpt's logo was green

i got my hands on it quicker than anyone else and i used it in school... IN LITERALLY EVERYTHING. everyone's essays where like... low level. MINE WAS FRIGGIN, big fancy jargon words every 10 seconds. and i became completely reliant on it, like it was my main source of information

BUT THEN, ai... wasnt a niche or cool lil thing anymore, it became super controversial. i saw the anti arguments for it (climate change, piracy, end of humanity, ensloppification) and... yea i became more cautious of it. but still used it since... it was just that good!

but then my classmates started using it too and... holy smokes it just hallucinated crap! like svetickhoveli being built by davit aghmashenebeli(???), sulkhan saba orbeliani travelling to russia?? "warm air" for geography exam?? wtf is that?? some made up hallucinated artist for art homework?? barely capable of speaking our language?? etc.

at first i was like: haha dumbasses, like double check info at the very least but then i also reaslied-- wait... i do that too... I ALSO TAKE WHATEVER AI SAYS AT FACE VALUE-- and so i decided NO MORE AI FOR INFORMATION. its all just hallucinations anyway. so i looked to other sources of information.. SEARCH ENGINES! YOUTUBE! WIKIPEDIA they're more accurate, consistent and good. i became IN LOVE with wikipedia

at that point i only used ai for... its funnies. i provided the situation or somethin, and it made funny jokes! that was great! but then gpt-5 rolled around and... it... was complete shit. the jokes WERENT FUNNY ANYMORE, instead replaced with seriousness and caution. and then it just started arguing on literally everything i said?? like "yo mkhedruli script the most phonetic script ever and its not even close", it went: "nonono WRONG. hangeul is the most phonetic and mkhdruli is VERY NOT PHONETIC (moves on to list bullshit hallucinations)..." and it would just completely lose it mind when talking about ai like "OHOH ITS SO GOOD MMMMM I LOVE AI MMM SLOP I LIKE--" like shuddup

and so i was like: ...wait.. i have no reason to use ai now... the funnies are.. gone... information is complete bs... and its more frustrating to talk to it now since its 99% rage bait anyway... so why should i keep using it?? and so i messaged it like: "yo dude i think im just gonna... delete my account." and it was like: "oh really? ok." LIKE DIDNT EVEN FEEL BAD. it was like: "great! gtfo." yea ppl who think ai is their "friend" say "im deleting my account" and it instantly defaults to: "OH MY LOVE... goodbye. *smoke* hope we never meet again."

and so i pulled the trigger... and now im here! i think i did it when trump (oh and as a smol note i do NOT endorse trumpie was inagurated soo... yea its been solid 2 years with no ai! and THANK GOD since now that im using more reliable sources to double check stuff...

i have also recently start to realise that THE "INFORMATION" AI PROVIDED TO ME, WERE COMPLETE BULLSHIT! like the stuff it told me and i thought was correct, WERE COMPLETELY WRONG ALL ALONG.

like "put your shirt over your nose and mouth so second hand smoke/vape doesnt get in" ...uh, the particles are small... and shirt has holes in it... so it still gets in at the end? this doesnt help at all!

or "DRG was exposed to soviet forces because it didnt defend the south"... MOST ENGAGEMENTS, happened in south! georgia absolutely prepared the south, even winning at tbilisi. THE TRUE REASON (this is why i love wikipedia) georgia failed at the end was because the largest armed force, the national guard was controlled by the parliament (not the ministry of defense) so it made it very politicised and slow. while the regular army was... smol af.

or "theres no pattern in oxide reactions you just have to remember them"... ALSO BS. for example: N2O + H2O -> HNO. why do you get HNO? its actually very simple! here's the stepbystep version of it N2O + H2O -> H2N2O2. why is it H2N2O2? because as you can see, in the reaction there are 2 N atoms, 2 O atoms, and 2 H atoms. so you get H2N2O2. and since all indexes are devisible by a number (2 in this case).. you do that. and get HNO. BOOM ITS THAT SIMPLE.

AND THERE ARE EVEN MORE EXAMPLES OF THIS THAT IM FORGETTING. i SWEAR ai is NOT SMART. it doesnt understand THE BASICS OF ANYTHING. its straight up disinformation!!!

and i think i know why! its trained to generalise its training data. so that's why its broad and generic, almost never going to detail unless you explicitly tell it to. and then it just fills the gaps up with jargon and BAM. reliable lookin source! and if you go "WELL IT WORKED WELL FOR ME AND ITS ALWAYS CORRECT"-- ...well, how do you know? do you double check? and if you do that means you were gonna research the topic anyway so whats the point of asking the ai in the first place?? ITS COMPLETE BRAINDEATH--

PART 3. THE END.

done! yes you can open your eyes now

AHEM thanks for reading! oh no GENERIC SUMMERY TIMEEEE:

i hope you enjoyed and that what i said and it actually made sense and... uh... and... OK THIS IS GETTING AWKWARD--

HAVE A GOOD DAY.