A dataset with only data from before 2024 will soon be worth billions.

blahyawnblah · 2026-01-08T15:54:06 1767887646

noir_lord · 2026-01-08T15:56:26 1767887786

I’ve already gotten into the habit of sticking “before:2022” in YT if what I’m looking for doesn’t need to be recent.

The AI slop/astroturfing of YT is near complete.

ares623 · 2026-01-09T00:17:56 1767917876

I would say around 2023. I refuse to believe the slop propagated that fast.

And there's more than enough content for one person to consume. Very little reason to consume content newer than 2023.

cbm-vic-20 · 2026-01-08T16:06:39 1767888399

Workaccount2 · 2026-01-08T22:53:40 1767912820

Sythentic data is already being embraced. Turns out you actually can create good training data with these models.