this post was submitted on 25 Mar 2026
401 points (98.3% liked)
Not The Onion
20984 readers
1832 users here now
Welcome
We're not The Onion! Not affiliated with them in any way! Not operated by them in any way! All the news here is real!
The Rules
Posts must be:
- Links to news stories from...
- ...credible sources, with...
- ...their original headlines, that...
- ...would make people who see the headline think, “That has got to be a story from The Onion, America’s Finest News Source.”
Please also avoid duplicates.
Comments and post content must abide by the server rules for Lemmy.world and generally abstain from trollish, bigoted, ableist, or otherwise disruptive behavior that makes this community less fun for everyone.
And that’s basically it!
founded 2 years ago
MODERATORS
you are viewing a single comment's thread
view the rest of the comments
view the rest of the comments
I hope that, at least, AI there will be using RAG system and some real information sources. Otherwise, as you have mentioned, it would be a hell of education.
What is RAG?
It's a chat bot that googles your question before answering in the hopes to cut down on hallucinations. It doesn't solve this problem at all.
Your explanation is not completely correct. More correct explanation would be: an AI chatbot that has an ability to gather relatable info to the user input from internal or external sources allowing the AI model to answer more precisely on questions even if the model wasn't trained on this data at all. This lowers the amount and degree of hallucinations to some point but doesn't eliminate them.
So... a search bot.
Basically, yes, but more nerfed. This ability is essential for many current AI chatbots and works on basic level of accessing specific needed info from storage. Easy example for a search bot will a chatbot that answers questions. Instead of giving LLM AI model access to all database(thousands of sentences), special algorithm(usually a semantic search) retrieves only a small relevant to the user input part of info(like only a dozen of sentences) and gives it to LLM model for further analyzing and formatting. This approach reduces the amount of tokens for which you pay, makes answers more correct and fast.
For a bot to be a true search bot such searching algorithm should be able to use internet and local storage together. And the interesting part is that you don't even need to have an LLM in this chain. You can only use RAG system to search for needed info in any text file. The only issue is that it will not be formatted and analyzed automatically.
I don't think it will make enough difference, but RAG stands for Retrieval Augmented Generation.
There's a few ways to do it, but basically it's a way add extra information to the conversation. By default the model only knows what it generates, plus what is in the conversation. RAG adds extra information to the mix.
The simplest approach is to scan the conversation for keywords and add information based on them.
So you ask "what is the capital of France" and instead of the model answering/hallucinating by itself, your app could send the full Wikipedia page for France along with your question, and the model will almost always return the correct answer from the Wikipedia page and hallucinate much less. In practice it gets a lot more complicated and I'm not up to date on recent RAG but the idea is the same.
Ah, so much better than our current systems the use significantly fewer resources for consistently correct answers.
A separate subsystem for an AI chatbot that allows it to get related to the user input information from text files(database) without scanning it all each time or having as an input to the promt, thus reducing hallucinations since instead of telling you something "from the head" it has an input in the form like this: user_input+info_content+memory.
Despite RAG being really helpful in many ways it doesn't eliminate hallucinations completely. Only lowers them to some point.