The Yes, This Big Header Banner
No stickies found.
Random Comment
Comment #3722
i've been on n for a while. got the same router as dad i think. 40 bucks, well worth it. though the firmware is different and i can't remote desktop or ssh into my home network anymore.

The wonders of AI, a Google and Reddit Story

at 15:31 - 4th, June 2024
Google recently paid Reddit 60mil to have their AI (LLM) go through the entire Reddit site so that it could learn and become more useful.
Open AI did this with X after Elon bought Twitter and he sued them for going through without permission.

On a slight tangent, this is the probably the first and only way that Reddit will make money since going public.

Anyway, knowing Reddit, you can imagine how smart this "AI" is now. A perfect example is someone asked google how to get the cheese to stick to pizza. It answered with, use glue.

Why did it give that answer? Because some dude 10 years ago joked about that's how he got his cheeze to stick to his cold pizza.

Fucking brilliant


Similar posts

Hey __________, Watcha Playin?
Yea i found a couple new hey ash watcha playing, but that would be for another post. this...
Rain in St. Kitts
Well, it's about time. since I came in February it has barely rained. Vikie can vouch, we...
Simcity 2013 on Mac is "Worse than the PC re
So we've come full circle. Simcity was released on mac this week and though it seemed lik...

Cocaine in Red Bull
Wow what a great way of getting high: German states ban Red Bull after discovery of coc...
My E3 Highlights
(media inside)
Certain Apps on Mac App store cracked already
I have several issues with buying shit off Steam, apple's istore for the iphone/ipad and ...

js js
News comment 1 | User comment 1052 | 22:11 - 6th, Jun 2024

There's a lot of information on reddit, and a lot of it is bad lol. They need to make models that can figure out how to rate the quality of the information and understand meme answers if they want to train an algorithm on a site like reddit. For MLM, it's garbage in garbage out otherwise.

What seems to be the issue is how much information is needed to train the models. They have run out of written content on some of them and are now transcribing youtube videos to make more content to feed them. Seems like the models take too much information to really be able to understand the content.

Alex Alex
News comment 2 | User comment 4968 | 14:14 - 7th, Jun 2024

God, if they're using the autogenerated close captions from youtube for the videos, we're in some serious trouble.

Join the conversation

Don't have a username? Register Now
Can't remeber your login? Find Password