What a garbage dataset to train off. The majority of everything there is all bots and AI anyway lol

Come on over to Mastodon, the water's nice

Dude, that’s gonna be one hell of a racist AI just based on the dataset it’s gonna receive

Wait this whole article just baselessly assumes that 1 new bluesky account = 1 person leaving twitter. That is so obviously unrealistic. Sure some people were probably curious and wanted to check out something new but that doesn't mean they will immediately switch platforms.

You can’t just make fun of Those Guys for endlessly believing fake bullshit while unquestionably parroting this garbage.

Just helpful to remember, as all this horseshit persists, Bluesky is Jack Dorsey. Dorsey is former head of Twitter and musk's good buddy.

It's all a losing proposition, whatever direction you move - except if you move... Away.

These mfs will use literally anything except for open source, decentralised social meda

Social media like Twitter preys on people's fear of missing out.

Manufactured Problem - not being up to date on internet bullshit. Marketed solution - be on twitter. Supply and demand died a long time ago. Now it's all about manufactured problems, and conveniently marketed solutions.

This sort of psychological manipulation in marketing works. It's why it's so hard to get people to leave websites like Twitter, reddit, fb ect. They've made their brands synonymous with media trends.

It took long enough, but I'm genuinely happy to see folks wising up and realizing they don't "need" Twitter. It's like watching a bunch of people break up with a toxic ex all at the same time.

So let's say on Twitter, someone blocks me and I can't read their post. Can't I just log out and read their post that way? I don't have a Twitter account, so I've never seen a blocked link before.

Opt out is such a shitty practice. Especially for AI generation

While anything that gets people off Twitter is good, I'm sorely unimpressed by those artists who "had to" to patronize the racist transphobic neo-Nazi hellhole "because my audience is there"... until Musk's policies happened to offend their own personal interests, by requiring training for their AI. Countless models trained on all public images already exist, jumping ship won't prevent their work from being scraped elsewhere, and frankly, any one image or even portfolio will contribute virtually nothing to the result, so quitting in protest is largely symbolic. But so many peoples drew the line at that, and not at Musk making "cis" a slur, or protecting child pornographers, or boosting white supremacist supremacy theories. It's really disappointing to see.

And that’s on top of all the Brazil users who fled to BlueSky when he refused to comply with a court order (and pay a fine) so they blocked Twitter for a few days. I’m not sure how many went back after he paid the fine but BlueSky was fairly popular in Brazil even during the closed beta so I’m sure a ton stuck around.

How do we know this bluesky isn't just the same shit run by different assholes?

lol

Lol, that's probably 30% of real people still using it

has so far only been used to train Musk’s own Grok AI

The Grok models are a laughing stock in the LLM space. They aren't good over APIs, and they're even less useful after Elon "open sources" them far later. Qwen 72B, and heck, Qwen 32B is already better than Grok 2, which is probably hundreds of billions of parameters. Qwen is runnable locally right now, Apache 2.0, and released day one. Grok 1 is... well, I dunno, no one has even bothered to try hosting it for anything.

I dunno what Twitter is doing with all those H100s Elon hoarded, but it seems like a big waste so far. Its certainly nothing to help the open source/self hosting space or to "decensor" and "democratize" LLMs like Elon fans seem to think.

Is 500,000 actually a lot? Honest question, my sense of scale with these things is shaky.

Bluesky is pretty great once you learn how to use it. I plan to slowly back off twitter.