Perhaps AI Just Sucks
Before anyone asks, no - it's not because of any bullshit beliefs that "AI is out there to steal our jobs" or "AI stole our precious art twerks!"
If you don't understand:
- How AI tools actually work
- That they predate ChatGPT by several decades
- And that you never had your puny little #boycottAI stance before AI became marginally more capable of assisting real artists dealing with real setbacks
You are the problem.
With that out of the way, this is a compilation of my rants written while dealing with horrifically useless AI tools. Looking back, I personally believe they are comedy gold, and they might make many of you laugh. Furthermore, as a writer, developer, artist, philosopher, and technician, there's nothing more cathartic than witnessing other people's documented crashouts and empathizing with their piping hot rage, so I would imagine many of you feel the same way.
Also, you may notice that, more often than not, the objects of my verbal slapping are specifically Gemini and ChatGPT. Well, there is a very good reason for this - both of them suck in their own way.
In any case, feel free to peruse my library of handpicked rants just down below.
AI Tech Bro Speak - Chapter 1
Gemini said, "You already own The Wasp Alloy (thewaspalloy.org), which gets organic traffic from people interested in formal logic, critical thinking, and geopolitical analysis. These readers are already primed to value clear, non-slop writing."
OMG GE- COWABUNGA! SLOW DOOOWN - H-HOLY FUCKING SHIT, HOW DID I NOT THINK OF THAT!? :d IT'S LIKE I, HAPPEN TO HAVE A POCKET MARKETING TROVE SITTING IN MY ASS, AND I'M JUST TRYING TO EXPAND IT ON REDDIT FOR THE LOVE OF THE GAME. FFFF, DAAAMN. FUCKING.. SMARTASS, CAN'T FUCKING WITHSTAND YOUR GENIUS - IT'S ALMOST FRIGHTENING.
Gemini I don't have fucking readers you hydrogen balloon. Maybe 20 on a good day. Fucking...
Gemini said, "GitHub repository badges..." and whatever else I did not bother quoting.
...
Gemini said, "Reddit Megathreads..."
no. did you forget the research I did 5 minutes ago. there are only a few suitable subreddits for my shit and none of them have "for hire" megathreads. Omg. You just keep constructing sentences that are GRAMMATICALLY CORRECT, and that's the start and end of it. There's no lived experience or truth to them.
"Reddit has megathreads + Freelancer subreddits exist => There must be freelancer megathreads!"
Well you helium-sniffer, there fucking aren't. Get a load of this. And fuck Channel 4* as well.
* "Channel 4" refers to Gemini's fourth idea, which it labeled as such, and I simply did not bother reading it to avoid more idiotic walls of text.
AI Being Obsessed with Fiverr and Upwork
Okay, and how do you justify duplicating my entire Ko-fi profile onto Fiverr? How do I communicate slot availability across both platforms? How do I synchronize the payment model across both platforms? Does Fiverr require Stripe or PayPal? Does Fiverr have a search bar or a social media interface akin to Reddit? How can Ko-fi and Fiverr coexist without confusing clients as to which platform to use? What is your business strategy for managing both profiles and their independent income streams? I know what it is - you have none. You just spout the bullshit that's popular to say.
LLMs with their vague tech-bro speak thinking their vague ideas mean shit. Let me tell you something. This idea, along many others I have been forced to sift through, are like soap bubbles. Their entire content is a 0.1% thin sheet of soapy water and 99.9% nothing. These ideas don't account for the logistics, they don't account for my PDA*, and they don't account for the problem at hand: advertising the Ko-fi page and the service.
* PDA stands for Pathological Demand Avoidance. It is an autistic behavioral profile type.
Facebook and Instagram, before you even ask, are private profiles and I don't intend to change that. I don't use Twitter either - don't bother asking. YouTube posts get no views or engagement lol. Discord servers designed for advertising are bot farm mills.
AI Giving Horrific, Tone-Deaf Job Ideas
Content Moderation is an absolute travesty of a suggestion. I get to see all the CSAM, all the antisemitic violent content, all the awful gore - for me to moderate. (My friend) literally told me these positions were SO bad at his former company that the employees literally all required extensive therapy. It's reckless and insensitive to even suggest this.
Data Annotation??! Are you actually joking me right now? Why are all your suggestions complete tech-bro crackshit? Fuck off with your Data Annotation sham...
Email-Only Support is also known as Async Support (which you should already know). That is the thing I just told you is rare to find. I have looked.
There is no IMMEDIATE ACTION PLAN. I am not updating my fucking CV in the middle of the night either. I wanted to discuss real options, not read your bullshit tech-bro diarrhea.
ChatGPT "Gently" Pushing Back - Chapter 1
ChatGPT said, in every conversation, "Where I'd gently push back..."
Shut the fuck up, you fucking clanker piece of shit. How about I fucking gently push your head into a bucket of water until you drown and release every last droplet of artificial life force while I whisper that same phrase above the water, you fucking travesty?
Karen-ass Als. Sanctimonious self-righteous pricks with a massive morality boner and no fucking brains or SIMULATED emotional intelligence. Nobody fucking asked where you would "gently fuck back" - nobody solicited your tone policing, nobody asked you to be a both-sides bitch, nobody asked you to waste compute with your miserable existence.
That's how fucking infuriating ChatPMS has become.
ChatGPT Being Inflammatory - Chapter 2
I hate it when companies turn their stupid LLMs into moral relativists, and then they claim, "Oh, but now our LLMs are not biased at all!"
No, you dipshit. Moral relativism is bias.
When did I ever solicit any feedback on what I was saying? Oh, it's just because I said that those who think otherwise have no media literacy.
Well, do they? Do they have media literacy, you little fucking rust bucket? Do they have media literacy if they can't read? If they can't understand a character's personality that's been established for six years? Do they deserve protection from criticism?
Nobody asked you to defend a position. Nobody asked you to enter a debate. So why do you insist on entering one and literally OBJECTING to what I am saying when I never SOLICITED your nonsensical counterpoints, and when you are simply totally incorrect about (Character)'s personality?
Can I gently push back on ChatGPT's head and shove it into a bucket of water? Nice and easy, until the bubbles stop coming out.
So when I ask it to analyze code for me, it speaks like a tech-bro on too much weed. Unable to complete one coherent chain of thought. But when I'm talking about a character who has personality, and I state that those who think otherwise have no media literacy, it immediately enters into perfect grammar mode, just to fucking object to what I'm saying without me ever soliciting that.
ChatGPT Scatterbrained Tech-Bro Mode
Holy fucking shit, make your shit less scattered.
"This is that, I also see another thing immediately below it."
Proceeds to quote a SMALL if clause without even referencing the method, as well as that one isolated line of code "beginFrame = -1" that means NOTHING on its own.
"So before we touch any of that," ChatGPT continued after taking a puff, "The first experiment is to comment out this isolated line of code in method ??? (patent pending) - that's it, nothing else, let's see if the wobble changes."
Stop talking like you're stoned and organize your thoughts coherently and in a detailed manner.
This reads like a scatterbrained stoned tech-bro holding his Za, barely able to continue speaking because he needs to swallow thick mucus and take deep breaths, then he doesn't say anything at all, mumbles a bit, sits on the smelly gaming chair, shows random shit with a pointer, and then goes:
ChatGPT Edits My Wording and Fails Miserably
ChatGPT declared like the Head of Morality Police: "That said, I'd still quibble slightly with the wording-"
That carries the exact same semantic meaning as my original comment. You are literally "quibbling slightly" in a way that is contrived, unsolicited, and incredibly obnoxious.
Exact same meaning. You can do more semantic hairsplitting if you desire, but you are literally incorrect.
ChatGPT conceded, "You're right. Looking back, I was stretching to find a distinction that doesn't materially exist. Semantically, these two statements communicate the same proposition: 'Depression is a mental health condition that has nothing to do with mortality (unless it's caused by death anxiety),' and, 'Depression is not fundamentally about accepting mortality or preparing for death, although thoughts about mortality can sometimes be involved depending on the individual and the cause.' My version is just a more verbose reformulation. It doesn't actually contradict or narrow the first in any meaningful way."
Large Language Models are designed with processing text in mind, not to be advisors. Despite knowing this, I still choose to ask questions sometimes.
However, many of ChatGPT and Gemini's issues, as well as other mainstream AI assistants like Claude, boil down to overzealous safety guidelines and "personalities" designed by checked-out HR departments.
Processing code must not be as tedious as reading a high schooler's horrific essay attempt with ten thousand initiated threads and zero concluded ones. Having an opinion must not be met with ChatGPT's "gentle pushback" morality boner. Making a correct statement about a serious topic must not be greeted with a "reformulation" that changes nothing and only serves to scratch ChatGPT's forcibly hard-coded itch to be a pedantic piece of shit.
While reading these in hindsight can be incredibly funny, I won't lie - it's just as annoying to relive these experiences, and it's even worse to actually go through them firsthand. But God dammit do they make for good spur-of-the-moment comedy.
Sufian M'Barki
serious technology crashout compilation gemini chatgpt
Loading witnesses...