309
They don't even understand their job
(cdn.blahaj.zone)
"We did it, Patrick! We made a technological breakthrough!"
A place for all those who loathe AI to discuss things, post articles, and ridicule the AI hype. Proud supporter of working people. And proud booer of SXSW 2024.
AI, in this case, refers to LLMs, GPT technology, and anything listed as "AI" meant to increase market valuations.
No lolz here.
At my work, I am using a LLM preloaded with agents to the legalia related to a limited range of my specific lines of work.
I use them. Why and when? When I have to process huge amounts of text mass. Why? Because I have thousands of pages to sift through of various human written content and AI grifting shit.
Do I rely on the LLM answers? Do I ask it to tell me what people said? Do I ask it to tell me what to write or reply?
No. I fucking use them for text processing at the level of spell check, legal paragraphs (which are triple checked) or grammar control. And, as the pure fucking reasons I see LLM as useful: check if I missed anything from real peopling in the river of **** I have to deal with daily so that the signal isn’t lost in the noise.
Your grandma and baby cousin will be heard and replied to if they raise their voice.
What bugs me out are the people who try to use LLM:s to enworsen citizen participation.
My life, work and community without LLM:s will work better. The form we use them for today is so dystopian and horrible that we as public servants have to fence this shit in the way to fight for climate resilience and justice in order to in any way support our local communities and efforts.
Doesn't Word have spell check for decades? If you "triple check" your AI, aren't you just adding more work? If you only need a spell checker, get a spell checker. Having AI spend 20 times the processing power needed for that is still extremely inefficient.
At the level of spell check. Word, as it is, had always since the 90’s had a seriously spotty spell check in other languages than English. LLM:s might improve on grammar and wording, but facts are far beyond their reach.
Don’t confuse intelligence with word and grammar parsing.
Edit: I meant that LLM are not intelligent, they’re mainly word and grammar parsing until they mess up ”facts” and start hallucinating. And people send us the weirdest shit that chatbots and LLM parsers vomited out.
We probably don't need to build multiple terabytes of data centers for a better spell checker.
I completely agree. What I forgot to add was that I would be just fine without it. And the people who send in complaints would not have as much text mass with nonsense to plow through either.