This is… really not true at all.

LLMs differ from humans in a very very important way when it comes to language: we know the meanings of the words we use. LLMs do not “know” things, are unconcerned with “meanings”, and thus cannot be said to be “using” words in any meaningful way.

Having thoughts isn’t an issue. The issue is engaging with the thought.

You’re sitting there meditating. Then your head goes “hey, I gotta pay my insurance bill”

And then you go, “oh yeah, the insurance bill. For my car. Do I need to change the oil yet? Oh no, I have another like 800 miles to go. What’s 800 miles from here? Cleveland? Man, that LeBron really has some longevity”

Or instead, you could go “I’m meditating”. No judgment though, just point it out. Just note that a thought occurred and redirect your attention to the fact that you’re meditating and to your breath or whatever else. Another thought will come up shortly thereafter. You can just redirect again.

“What if Chromium had its own fake money?”

Then you don’t belong in a discussion forum.

If you’re gonna post something wrong and stupid, and then when someone goes through the trouble of correcting you your reaction is “oh fuck you, I don’t care”, then write your opinions in a journal and bury it in your backyard.

I started playing Factorio about a month ago and I’m completely addicted. I love it so much. I guess there’s not much more I can say about it, given how everybody but me had already heard of and played this game. But it’s definitely my #1 gaming highlight of the year.

“Firefish isn’t just a cool name”

Yeah, it’s also a generic name that isn’t very good?

The deck of cards thing is a truly insane stat. It’s not just more unique shuffles than grains of sand on Earth; there are more unique ways to shuffle a deck of cards than there are atoms in the Milky Way.

If the rumor is true that OpenAI is using libgen to obtain books, then this will be a very interesting fight.

Authors profiteering from arcane copyright laws vs. a sleazy company that hypes up an LLM as if it were HAL from 2001. Who is worse? Who should lose?! I’m on the edge of my seat already!

Yes! …well, with a caveat.

Huggingface models (even the ones published to the platform by Google or fb) are available for anyone to use, but they’re not quite intended to be chatbots; they’re more like pretrained machine learning models for data science pipelines.

Still, if you know python or are willing to learn, you can use huggingface however you want. You’re totally allowed to download a model, open it in python, and ask questions to it. Or feed it a long text and ask for summaries etc.

huggingface is full of models not created by a large corporation. Maybe they mean those?

No it doesn’t. There is no contradiction there. You’re talking nonsense.

Do you know how Goebbels infamously described the Weimar Republic’s attitude towards the Nazi Party?

Essentially, if you’re not familiar, Goebbels was quoted as saying something along the lines of “I have no idea why they allowed us to exist in their system and take power. It should have been obvious that as soon as we gained power, we’d use that power to dismantle the system and ban the traditional parties from politics. Anyone with half a brain would have realized that and banned us first.” Then he cackled like a witch.

Ok I made up the part about the cackle.

I’m learning to play the talkbox. It’s tons of fun and it’s totally a valid way to make up for the fact that I can’t sing my way out of a paper bag, lol.

Lol, twitter can’t even pay for their servers. They can’t afford a lawsuit!

This is one of those “oh duh, no shit” kind of things where in reality… yeah, that probably does need to be said out loud. So fucked up.

So the one thing on Reddit that you wish to leave behind is mods deleting transphobic comments? Lol