AI

God, I'm fucking sick of Generative AI and large language models. EVERYTHING wrong with them aside (and the list is extremely long), they are being constantly forced down my throat, and not ONCE have they EVER been useful in any appreciable way. Yeah, yeah, whatever, SOME people are helped by LLMs. I'm sure there are specific and boring and unflashy situations where LLMs are helpful. OpenAI is losing billions of dollars a year, I truly don't give a fuck. Anyway, I'm like, the poster child for someone who Sam Fuckman thinks would benefit from AI:

Unfortunately for Shit Fartman and his cronies at OpenASS, I have a few other qualities:

And every stupid fucking day of my life, my workplace is trying to force AI down my throat. Here's how it has been so wonderfully helpful to me.

First Contact

Because my work uses Windows Edge for some godforsaken reason, Copilot kept trying to answer my google searches. I'm not saying bing. I thought hey, well, what if it's useful? So I googled a question I had: Does ResearchRabbit actually use AI? Cause their url is researchrabbit.ai, and people use it as an example of AI being useful. (Side note: I'm pretty confident it doesn't. It's a literature mapping software tool, and the literature has already been mapped by someone else. Did the someone else use AI? Yeah, if by AI we mean machine learning and information retrieval, but Research Rabbit didn't. I'm pretty sure just a pretty shell on top of other third party databases like Semantic Scholar.) So I google 'does ResearchRabbit really use AI?'. Copilot's response? To describe how I can use ResearchRabbit and rattling off their sales pitch. NOT WHAT I FUCKING ASKED. So I found out by doing my own damn research, reading web pages like a normal human being.

I gave it one more chance after this. At work we use a very poorly documented student management system, so I googled how to do a certain action in the software. Copilot didn't have shit to say about that. Turns out when your LLM depends on huge amounts of information, precise queries about uncommon topics pose a problem.

LLMS ARE BEING SHOVED DOWN MY FUCKING THROAT

My boss uses them. We had a whole workshop about how we could use them. Here are some highlights from the workshop.

Actual Application

Okay, fuck it. I'll try it. It makes me feel sick every time I do because I visualise the water being wasted, but I'll try it once. I write reports, but I'm too concise, so I tell ChatGPT (running through duck.ai cause I'm not letting those fucks train on my inputs) 'Rewrite the following sentence to be longer'. ChatGPT gamely pukes up several large paragraphs of useless dogshit, because it has made up bullshit I didn't say. Instead of, I don't know, rewording what I DID say to be longer. Like, instead of 'the students like the library', it could say 'students indicated they appreciated using the library'. But instead it's like, 'students extremely appreciated the library, describing it as warm and welcoming'. NO THEY DIDN'T!!!! SHUT THE FUCK UP!!!!!

Fuck, man. I'm sick of this shit. It's being pushed at me constantly, and even when I try to engage on its own terms, it's utterly useless to me. And yet OpenASSHOLE is using MY web footprint to waste hundreds of litres of water a day, create extremely damaging noise pollution, waste silica and give me a headache. Fuck you.