BrikoX
TechnologyAI-generated dossiers from the Jmail team.
Archived version: https://archive.is/newest/https://www.theverge.com/policy/879508/jikipedia-epstein-email-encyclopedia
- Highly approve as a long-time Wikipedia editor, though I wish they'd write it without LLM generation.
- "Jiki" sounds like an ethnic slur, and I have no idea why.
Edit: I take it back.
The Jmail X account said that they’ll be implementing the ability for users to report inaccuracies and request changes soon.
That's not what a Wiki is. Although I guess they're not technically claiming to be one; they're a Jiki.
Now that's good jiki!
Those damn dirty jikis. They're ruining pedophilia for all of us!
Someone should put together a deck of cards and sell them. Kinda like what they had in Iraq.
Some guy on tiktok tried that with billionaires and every major entity he had to go through to be able to produce the cards and sell them treated him like a domestic terrorist and shut him down
I bet some Chinese manufacturers wouldn't object.
I dunno why but the head-catching basket is kinda cute. Like it could have been a wooden or metal box but it's this dainty looking woven basket. Love it.
Here's the link for the lazy: https://jmail.world/wiki
I tried the AI search by the way and it sucked. The ONE thing AI could do spectacularly well is to search through vast troves of documents for relevant terms.
I think you're thinking of a search engine, AIs are bad at searching. They just make up a response related to the search term based on their training data.
AI systems designed for searching normally use https://en.wikipedia.org/wiki/Retrieval-augmented_generation afaik, which involves directly referencing documents rather than guessing based on initial training data.
Wouldn’t you just train them on the documents to be searched?
You get a lot fewer hallucinations if it’s presenting data from sources rather than from its neural network alone. Training data isn’t, like, “in” the AI. It’s just used to shape its creation.
Normally no, because that is much more difficult, resource intensive, and harder to get reliable results than separately looking up the information and including it in the prompt.
And if you train them on the document trove, they will be able to answer questions about it. It is a straight up trivial task.
Make up answers about it. The answers might be right, or they might be wrong, you won't know unless you read the actual data. So helpful ..
"Give me the line numbers corresponding to the saudi sheik saying he liked the torture videos"
Are you trying to be obstinate on purpose?
I don't think I'm trying to be obstinate, I just am.
A natural! Mazel tov!
If we are talking about LLMs, the other commenter is entirely right about how they function. But I'm not sure you two are talking about the same technology.
Can an LLM provide me the information I want given a search term if trained on the given dataset? Yes. That is all.
It can provide you some information that looks similar to what you'd want. Whether it is correct is another question.
RAG can help to a degree but hallucinations still happen quite a bit.
Maybe
Oddly, there's no jikipedia article on Jeffrey Epstein himself yet.
