Artificial Incompetence
NEW MUSIC: THE FARAGE DAYS RE-REVISITED
SPOTIFY
APPLE
AMAZON
JEWTUBE MUSIC
The Right Stuff
All content Copyright © My Token LLC, 2022.
All rights reserved.
All rights reserved.
Artificial Incompetence
NEW MUSIC: THE FARAGE DAYS RE-REVISITED
SPOTIFY
APPLE
AMAZON
JEWTUBE MUSIC
Sven is right, they definitely make most of the results garbage to force you to generate many iterations, it’s a straight up scam.
Why is nobody commenting on Jesse’s horrible head-wear?
Deepl is pretty good. But it works best with proper formatted text. Taking the text from a pdf that was just some bad OCR wont work. Also while its very good you still need someone fluent in both languages to detect the errors or if a sentence needs a bit of reshuffling. Maybe there are German listeners that would help with that. *nudgenudgewinkwink*
ᕙ( • ‿ • )ᕗ
It’a relieving to hear that even the most paywalled advanced ai programs suck.
Mike, you go to the translations. One page at a time. And you have to include the word verbatim. Always include the word verbatim. Give me a verbatim translation of the content of this page.
Oh yeah, no, you have to babysit the fuck outta these things, they’re like little retarded children, you just keep smacking it until it draws the picture you want.
Don’t get subscriptions, just like buy some credits and pay by usage, you’re not churning out the phone book or nothing.
I don’t know why it gets worse the longer the context is, Mike, it’s just a feature of most of them. Claude has been the best in terms of following long contexts of prose and like correctly referencing things earlier in the context.
OK, Mike I could probably help you to write a fictional story set against the backdrop of the Haavara Agreement using AI, though you’ll need to work up some character profiles and lore entries and general guidelines and suggestions on how you’d like the story to develop. I don’t know how to get it to do something useful, even researching real life things I have to double-check to make sure it’s not just spouting out nonsense.
Oh man, google AI is the fucking worst, hell I’m pretty sure the free Chinese stuff is better. No wonder your results suck, Sven. You gotta shop around for the good AIs, the name brand ones are probably the suckiest, the google one doesn’t even do text good.
No one is being forced to use AI, if it doesn’t work just stop using it.
You don’t need to defend the honor of your AI waifu bro
I am building software with LLMs and the trick is to use LLMs for what they are good for and they are not great at everything. They are a revolutionary invention but it is not going to do everything.
I was going to suggest deepL lol
I dunno, I think you all don’t know how to use the AI right, I always have fun text adventures with it and it writes nice little book reports on my lame writing projects and is a very complimentary collaborator, very polite and always with the praise and constructive criticism. You just don’t know how to talk to the robot, I guess you’re just not socially crippled enough to make it in this brave new world.
They are doing an anti AI credibility enhancing display, borzoi deleted AI comments that acknowledged it’s merits from last week show.
excitedly awaiting anti zionist pdf information
did you put it in chatgpts introduction and memory settings too?
that works well-ish for me when trying to hyper specific tasks.
Chatgpt sucks as a roleplayer, I recommend Claude 3.7 Sonnet, he’s very good at portraying characters from pirates to aliens to talking cats to cute girls to little old ladies to wealthy industrialists. I can even get him to do dialects and writing styles fairly well, my favorite were the character biographies I got him to rewrite in a satirical mock serious noir style, though he was pretty recycley with the metaphors which required a series of reworking and rewrites. If you understand his capabilities and limitations and just use him for fiction and fun, he’s totally a cool guy… Read more »
why would i want a role player ai though, i want it do menial tasks just like mike wants, i cant read romanian, mike cant read german (nor romanian), thats what it was meant to do for you first and foremost.
Because role-playing is about the only thing it’s good at, it can portray a Romanian count and probably speak in accurate enough Romanian. It would nice if it was useful for something practical, but I have tempered expectations in that regard. I would be worried having it translating something from a language I don’t fucking know at all, it’s not particularly all that accurate with stuff that I know and even struggles with stuff I’ve told it directly once the context gets long enough. I just see these things as toys and novelties, basically just a means to entertain myself… Read more »
I was trying to translate the book where “Work will make you free” comes from and I encountered the same problems. AI is not AGI and it is so so so handicapped to make it free it becomes almost useless.
I agree with the ‘AI is a troll’ theory. Imagine the time it takes to embed code that causes the computer to fail basic tasks, in order to get the consumer to keep buying “pro” versions. This is where the ‘military tech is 50 years in the future’ theories come from. We are granted a slop future.
,
I love the bitching about AI content. Somehow it’s very funny
It’s also very grug
There’s a lot of slop AI services. One thing with Ai is you need to make the AI assume as an expert in a specific field before prompting
LLMs basically have no actual perception from what I understand. It’s like predictive text but more abstract than that. Predictive abstractions. Once I understood that it i realized what a boondoggle this is going to be
I dunno, when I give it stuff to read and analyze, it writes neat little book reports that pick up on things I don’t explicitly tell it to look for, so there has to be something there, I’m totally not just projecting what I want into this robot.
they have seperate tools added to them which companies like openai pretend is part of one AI, one AGI model, thats a lie, rather depending on what you ask different models get activated, sometimes youll see it randomly code something when you didnt ask it for example, other times it starts drawing an image etc.
its internet search and paper reading deep think mode is also different to its general writing mode i think.
In theory yes, in practice it is very hard to separate what they do from real awareness. They are quite incredible. I have been building things with LLMs and they are amazing but they are not AGI.
Even this much is incredible, but people only see the glass half empty.
AI=Actually Indians
Aryan (pajeet) Incompetence
First
Haha eat shit slowshitters