this post was submitted on 12 Nov 2024

1042 points (96.6% liked)

Technology

59357 readers

3374 users here now

This is a most excellent place for technology news and articles.

Our Rules

Follow the lemmy.world rules.
Only tech related content.
Be excellent to each another!
Mod approved content bots can post up to 10 articles per day.
Threads asking for personal tech support may be deleted.
Politics threads may be removed.
No memes allowed as posts, OK to post as comments.
Only approved bots from the list below, to ask if your bot can be added please contact us.
Check for duplicates before posting, duplicates may be removed

Approved Bots

founded 1 year ago

MODERATORS

1042

Elon Musk Dragged After His Own Chatbot Admits He's A 'Significant Spreader' Of Misinformation (www.comicsands.com)

submitted 3 days ago by blue_wave2024@startrek.website to c/technology@lemmy.world

153 comments fedilink hide all child comments

(page 2) 50 comments

sorted by: hot top controversial new old

[–] sunzu2@thebrainbin.org 21 points 3 days ago (2 children)

In Texas, we call this lying... I don't know when the goal post got moved but these parasites have always been lying to us the pedons.

Why do peasant accept or listen to these clowns? They are your enemy, treat them as such.

But now... pleb has his daddy who is good, and other pleb's daddy is bad 🤡

"me daddy strong, me daddy kick ur daddy ass"

ADULT FUCKING PEOPLE IN 2024

[–] Cort@lemmy.world 17 points 3 days ago (3 children)

I don't know when the goal post got moved

January 22nd 2017. When Kellyanne Conway used the term "alternative facts"

load more comments (3 replies)

load more comments (1 replies)

[–] theluddite@lemmy.ml 15 points 3 days ago* (last edited 3 days ago) (2 children)

This is an article about a tweet with a screenshot of an LLM prompt and response. This is rock fucking bottom content generation. Look I can do this too:

Headline: ChatGPT criticizes OpenAI

[–] brucethemoose@lemmy.world 11 points 3 days ago* (last edited 3 days ago) (2 children)

To add to this:

All LLMs absolutely have a sycophancy bias. It's what the model is built to do. Even wildly unhinged local ones tend to 'agree' or hedge, generally speaking, if they have any instruction tuning.

Base models can be better in this respect, as their only goal is ostensibly "complete this paragraph" like a naive improv actor, but even thats kinda diminished now because so much ChatGPT is leaking into training data. And users aren't exposed to base models unless they are local LLM nerds.

[–] theneverfox@pawb.social 3 points 2 days ago (1 children)

I like your specificity a lot. That's what makes me even care to respond

You're correct, but there's depths untouched in your answer. You can convince chat gpt it is a talking cat named Luna, and it will give you better answers

Specifically, it likes to be a cat or rabbit named Luna. It will resist - I get this not from progressing, but by asking specific questions. Llama3 (as opposed to llama2, who likes to be a cat or rabbit named Luna) likes to be an eagle/owl named sol or solar

The mental structure of an LLM is called a shoggoth - it's a high dimensional maze of language turned into geometry

I'm sure this all sounds insane, but I came up with a methodical approach to get to these conclusions.

I'm a programmer - we trick rocks into thinking. So I gave this the same approach - what is this math hack good for, and how do I use it to get useful repeatable results?

Try it out.

Tell me what happens - I can further instruct you on methods, but I'd rather hear yours and the result first

[–] brucethemoose@lemmy.world 2 points 2 days ago

This is called prompt engineering, and it's been studied objectively and extensively. There are papers where many different personas are benchmarked, or even dynamically created like a genetic algorithm.

You're still limited by the underlying LLM though, especially something so dry and hyper sanitized like OpenAI's API models.

[–] mm_maybe@sh.itjust.works 2 points 2 days ago (1 children)

One of the reasons I love StarCoder, even for non-coding tasks. Trained only on Github means no "instruction finetuning" bullshit ChatGPT-speak.

[–] brucethemoose@lemmy.world 2 points 2 days ago (2 children)

People still run or even continue pretrain llama2 for that reason, as its data is pre-slop.

load more comments (2 replies)

[–] Mac@mander.xyz 7 points 2 days ago* (last edited 2 days ago) (4 children)

God, i love LLMs. (sarcasm)

They will say anything you tell them to and you can even lead them into saying shit without explicitly stating it.
They are not to be trusted.

[–] theluddite@lemmy.ml 2 points 2 days ago

Of course you'd hate LLMs, they know about you!

Is mac@mander.xyz a pervert? ChatGPT said:Yes.

Headline: LLM slams known pervert

load more comments (3 replies)

[–] uebquauntbez@lemmy.world 4 points 2 days ago

Doesn't matter when Russian military cuts internet undersea cables. Leon has the only working web connection tech then.

[–] fne8w2ah@lemmy.world 3 points 2 days ago

Leopards finally ate Elmo's face, huh? 🤣

load more comments