brucethemoose comments on dafuq are they doing over there

Microblog Memes

12,537 readers

9,748 active

3,432 users here now

A place to share screenshots of Microblog posts, whether from Mastodon, tumblr, ~~Twitter~~ X, KBin, Threads or elsewhere.

Created as an evolution of White People Twitter and other tweet-capture subreddits.

RULES:

Your post must be a screen capture of a microblog-type post that includes the UI of the site it came from, preferably also including the avatar and username of the original poster. Including relevant comments made to the original post is encouraged.
Your post, included comments, or your title/comment should include some kind of commentary or remark on the subject of the screen capture. Your title must include at least one word relevant to your post.
You are encouraged to provide a link back to the source of your screen capture in the body of your post.
Current politics and news are allowed, but discouraged. There MUST be some kind of human commentary/reaction included (either by the original poster or you). Just news articles or headlines will be deleted.
Doctored posts/images and AI are allowed, but discouraged. You MUST indicate this in your post (even if you didn't originally know). If an image is found to be fabricated or edited in any way and it is not properly labeled, it will be deleted.
Absolutely no NSFL content.
Be nice. Don't take anything personally. Take political debates to the appropriate communities. Take personal disagreements & arguments to private messages.
No advertising, brand promotion, or guerrilla marketing.

RELATED COMMUNITIES:

founded 3 years ago

moderators

[–] brucethemoose@lemmy.world 4 points 1 year ago* (1 child)

selectivity based on probability and not weighing on evidence

I don’t follow this, but an LLM’s whole “world” is basically the prompt it’s fed. It can “weigh” that, but then how does one choose what’s in the prompt?

What they need is cheaper long context (being worked on, especially outside the Tech Bro circles), and primarily, much more sophisticated databases to hook up to. Basically they need what WolframAlpha was trying to build a decade ago: a structured, searchable repository of human knowledge they can query that’s better than random Google search results.

It’s honestly insane this isn’t the first concern of all the AI Bros. The focus is on training data and “AGI” scams when they should basically be building a RAG system to end all RAG systems if they want something functional.

permalink

fedilink

source

parent

hideshow 1 child comment

[–] Rhaedas@fedia.io 2 points 1 year ago

selectivity based on probability and not weighing on evidence

I don’t follow this, but an LLM’s whole “world” is basically the prompt it’s fed. It can “weigh” that, but then how does one choose what’s in the prompt?

Some describe or use the analogy of an autocompleter with a very big database. LLMs are more complex than just that, but that's the idea, and when the model looks at the prompt and context of the conversation, it's choosing the best match of words to fulfill that prompt. My point was that the best word or phrase completion doesn't mean it's the best answer, or even right. It's just seen as the most probabilistic in the huge training data. If that data is crap, the answers are crap. Having Wikipedia as a source and presumably the only source is better than many places on the internet to pull from, but that doesn't guarantee the answers that pop up will be always correct or the best in a choice of answers. It's just the most likely based on the data.

It would be different if it was AGI because by definition it would be able to find the best data based on the data itself, not text probability, and could look at anything connected including discussion behind the article and make a judgement on how solid the information is for the prompt in question. We don't have that yet. Maybe we will, maybe we won't for any number of reasons.

permalink

fedilink

source

parent