AI shovelmaker Nvidia has been looking into making its own AI models. To that end, it’s been vacuuming up YouTube videos like nobody’s business. According to leaked internal communications obtained…
Anyone remember when that LLM scraped and learned from Twitter but subsequently had to be taken down because twitter folks. This should be interesting.
Hell, I really hope it’s scraping the comments too.
Hell, I really hope it’s scraping the comments too.
With what an absolute fucking trash fire YouTube comments are at the moment and Google being suspiciously silent about it, I’m sure that won’t cause any issues at all.
I’ve found it varies extremely.
Community/channel/field seems to make a big difference in the usual kind you’ll see, but there appear to be platform problems ito blocking some stuff from being posted/caught, too
Tay was a tad different than simply scraping Twitter as it learned based off responses and there was a concerted effort to troll it and make it racist.
These things are scraping absolutely fucking anything that might get them closer to Superintelligencetm, I fully believe the comments are getting scraped as well.
Anyone remember when that LLM scraped and learned from Twitter but subsequently had to be taken down because twitter folks. This should be interesting.
Hell, I really hope it’s scraping the comments too.
do you mean Tay? that was a touch pre-current-LLM, but it was demonstrative nonetheless
That’s the one!
With what an absolute fucking trash fire YouTube comments are at the moment and Google being suspiciously silent about it, I’m sure that won’t cause any issues at all.
“Never interrupt your enemy when they are making a mistake” - Google, circa 2024
“I bent my Wookiee” - Google, circa 2024
tbh my experience has been that YT comments are dramatically better than they used to be
Really? Most I’m seeing lately is obvious bot spam with content that’s not only blatantly against the rules but outright disgusting.
CW: mentions of animal abuse, violence, and racial slurs
And that’s not even the worst ones. I’ve seen some which mention and promote CSAM. Reporting these seemingly does nothing.
holy shit. we are having completely opposite experiences
fucking hell -_-
that’s worse than the shitty ones I occasionally get to see
I’ve found it varies extremely. Community/channel/field seems to make a big difference in the usual kind you’ll see, but there appear to be platform problems ito blocking some stuff from being posted/caught, too
Generative AI animations already draw completely cursed faces*, I can’t wait to see what they do with youtuber face.
* Gee I wonder why Sora used a balloon head for their big demo. I’m sure that’s a coincidence.
Tay was a tad different than simply scraping Twitter as it learned based off responses and there was a concerted effort to troll it and make it racist.
I’d argue there is a decades old concerted effort to troll and make racist the entire world wide web, by very dedicated actors no less.
These things are scraping absolutely fucking anything that might get them closer to Superintelligencetm, I fully believe the comments are getting scraped as well.