I'd add that Shams being the biggest NBA reporter/insider without writing ability / insight, and him instead using his connections to break news fast is not unique to the NBA. In soccer, Fabrizio Romano is pretty much the biggest soccer reporter/insider-- if you follow the transfer market in the summer or January, he is the guy people look too because he breaks news pretty fast and pretty accurately, to the point where even teams have been annoyed with him for breaking news before there's even a hint it's happening. Like Shams, he doesn't really write any traditional articles, but he's big off his connections and ability to break news. I'd expect most sports to end up like this in the long-run if it isn't like that largely already.
Agreed - a lot of sports are experiencing this. I think that the 'breaking news' function of journalism and the 'in depth reporting/analysis' are just going to become two entirely different jobs.
Do you intend to cover youtube's move to shutdown adblock users? Seems similar to reddit's API changes, or maybe the precursors that led to that situation in the first place.
Honestly I forgot this happened because I use ublock and it's so effective that within a day I had youtube back to ad-free. Might be worth covering!
My instinct here is to say (without much forethought) that this isn't much like reddit's API changes. YouTube is a fully mature, profitable platform that *mostly* has a good relationship with its userbase. The adblock people who care will figure out a way around it, and those who don't care will just watch the ads.
Interesting. I use a mix of ublock, ABP, sponsorblock, and noscript, and youtube is able to detect something in there. I'll try turning some stuff off and see what happens, since using invidious / yewtu.be is a bit of a headache and loses a lot of the benefits of having an account.
One of the things I've seen is that using multiple services can make it worse. I use ublock only, I'd try that. But now I'm researching and will probably include it in the next weekend roundup!
> A new ‘data poisoning tool’ lets artists poison their digital art to prevent AI from using it. Essentially, the tool takes your art, changes a bunch of pixels in a way that are invisible to the human eye but legible to AI systems.
It doesn't really work though, and there's no way it could ever work in general.
This is an example of an "adversarial attack". Basically you can always find one against any ML model; it seems to be impossible to design one that can't be tricked into thinking a cat is a giraffe or something, with a picture that looks no different to humans.
But the same goes the other way round. If you poison all the images, someone can always make a new model that handles them fine. Basically whoever's newest has the advantage.
Some other attacks I've seen are pretty easy to remove from images too, like if you make a few variants with filters on them (called "augmentation") it cleans it off.
Re. ‘AI enshittifies Google’ and bad answers from googling stuff, wasn’t this the blog that introduced me to the ‘can you melt an egg’ fiasco from a couple of weeks ago?
My experience with Twitter's algorithm echoes Nate's. I have a separate burner account that I just used to play around with sometimes. All of the people I follow with it are either musicians or vaguely left-leaning pop culture podcasters, and without fail, I kept getting push notifications from conspiracy-minded, sorta anti-vax and election-denying posts. And it wasn't even a mix. The only accounts I got those notifications from were right-wing nutjobs and barstool dumbfucks, there was no, like, dirtbag left equivalent. It was very telling to me about the kind of stuff getting prioritized over there.
Here I thought 'Talcum X' was the canonical SK nickname and it doesn't even make it into the footnote!
Clearly mistakes have been made
There are just so many good ones
I'd add that Shams being the biggest NBA reporter/insider without writing ability / insight, and him instead using his connections to break news fast is not unique to the NBA. In soccer, Fabrizio Romano is pretty much the biggest soccer reporter/insider-- if you follow the transfer market in the summer or January, he is the guy people look too because he breaks news pretty fast and pretty accurately, to the point where even teams have been annoyed with him for breaking news before there's even a hint it's happening. Like Shams, he doesn't really write any traditional articles, but he's big off his connections and ability to break news. I'd expect most sports to end up like this in the long-run if it isn't like that largely already.
Agreed - a lot of sports are experiencing this. I think that the 'breaking news' function of journalism and the 'in depth reporting/analysis' are just going to become two entirely different jobs.
Cream Abdul-Jabbar is too good.
Do you intend to cover youtube's move to shutdown adblock users? Seems similar to reddit's API changes, or maybe the precursors that led to that situation in the first place.
Honestly I forgot this happened because I use ublock and it's so effective that within a day I had youtube back to ad-free. Might be worth covering!
My instinct here is to say (without much forethought) that this isn't much like reddit's API changes. YouTube is a fully mature, profitable platform that *mostly* has a good relationship with its userbase. The adblock people who care will figure out a way around it, and those who don't care will just watch the ads.
Interesting. I use a mix of ublock, ABP, sponsorblock, and noscript, and youtube is able to detect something in there. I'll try turning some stuff off and see what happens, since using invidious / yewtu.be is a bit of a headache and loses a lot of the benefits of having an account.
One of the things I've seen is that using multiple services can make it worse. I use ublock only, I'd try that. But now I'm researching and will probably include it in the next weekend roundup!
> A new ‘data poisoning tool’ lets artists poison their digital art to prevent AI from using it. Essentially, the tool takes your art, changes a bunch of pixels in a way that are invisible to the human eye but legible to AI systems.
It doesn't really work though, and there's no way it could ever work in general.
Can you elaborate on that? Their test run seemed convincing.
This is an example of an "adversarial attack". Basically you can always find one against any ML model; it seems to be impossible to design one that can't be tricked into thinking a cat is a giraffe or something, with a picture that looks no different to humans.
But the same goes the other way round. If you poison all the images, someone can always make a new model that handles them fine. Basically whoever's newest has the advantage.
Some other attacks I've seen are pretty easy to remove from images too, like if you make a few variants with filters on them (called "augmentation") it cleans it off.
Re. ‘AI enshittifies Google’ and bad answers from googling stuff, wasn’t this the blog that introduced me to the ‘can you melt an egg’ fiasco from a couple of weeks ago?
Gonna be honest with you, I can't remember all the stuff that goes into the weekly roundup posts every week. So it's a maybe?
No worries, if it was you then thank you! It amused me no end
My experience with Twitter's algorithm echoes Nate's. I have a separate burner account that I just used to play around with sometimes. All of the people I follow with it are either musicians or vaguely left-leaning pop culture podcasters, and without fail, I kept getting push notifications from conspiracy-minded, sorta anti-vax and election-denying posts. And it wasn't even a mix. The only accounts I got those notifications from were right-wing nutjobs and barstool dumbfucks, there was no, like, dirtbag left equivalent. It was very telling to me about the kind of stuff getting prioritized over there.