The real story of AI is gonna be the way a whole lot of people are gonna pretend that they “couldn’t help” and that their cruelty was the fault of folks warning them

And that the actual draw of AI wasn’t the tech but the dominion over undesirables it gave

And the ability to ignore inequality

One possibility is that they came up with this circumcision nonsense in part to have an explanation of why autism is diagnosed much more often in men than women. Just working backwards to get the answers they want.

Soo everytime America swears that something has never happened before

It has happened and usually to a Black person

History may not fully repeat but it rhymes with

Also to me, a long-time known Autistic person, it feels like they're trying to make an explanation for "why boys are more likely to be diagnosed with ASD than are girls".

And they've already decided acetaminophen/paracetamol causes Autistic people to exist, so.

There's already a segment of the population that believes circumcision is common in the US due to some sort of nefarious Jewish plot. I've never gotten a clear idea of WHAT that plot it supposed to be, but I've seen it said by antisemites a bunch of times over the years.

Lest anyone think this is just a joke, remember that some transphobes explicitly claim that trans people are a plot by the Jewish people to undermine white people. A lot of conspiracy stuff ultimately becomes "It's the Jews fault", and medical conspiracy theories are no exception.

Forgive me for being furious whenever my white colleagues seem to be able to commit all manner of journalistic wrongs and be rewarded with multiple income streams while I am left to wonder how the hell I'm going to pay my already exorbitant ACA premiums as I float my living expenses from my 401K

I was concerned because his lack of addressing any of this made it seem like he doesn't even see that it's a problem to have a system that discriminates and not try to address that. Which, to be fair, is a conclusion I may not have come to if I did not already distrust Bluesky T&S.

There are really no good options for this, manually reviewing images for violations on a large platform is actually a VERY tough job that takes a significant mental toll on people doing it due to all the messed up shit they have to see. Moderation tools are important, and they're difficult to design

I feel like I screwed up royally by not mentioning any of the complexities involved in trying to fix this problem when I posted this. I was struck by the casualness of it and the lack of addressing the obvious "What are you going to do to fix this?" questions, and I left out important details myself

My real problem isn't that the mistakes are happening, it's the way that the Head of Trust and Safety is treating the issue as though there is nothing that can be done about it without even attempting to explain what they are doing to mitigate it and/or why it is difficult to solve.

I feel a made a major mistake in not mentioning one thing here: there may not be a good way to deal with this, as it is a problem with the model itself and thus there is only so much Bluesky can do, and the company making the model itself is probably trying to reduce such false positives.

y'all ever seen human centipede?

Can you clarify, is this explicitly stated in writing in the contract Bluesky has with them that they are not allowed to retain or train on the data?

Now I'm just imagining Nintendo making a vaguely soulslike Mario spinoff, where you get outfits/powerups instead of equipment and use jumping instead of dodge rolling.

Not gonna lie, I think it could be awesome depending on execution.

.... the website he links to shows that this company also works on generative ai. I want to know what assurances, if any, there are that this company is not scraping every image posted to this site for use in their ai models without affirmative consent of the users.

I really want to point out that this is the HEAD OF TRUST AND SAFETY casually saying that their automoderation tools falsely flags depictions of disabled people as adult content with regularity. So this is a known issue, and he gives no indication they are attempting to stop this.

If this is a known issue, your automated system has an ableist problem you have known about and did nothing to stop for how long?

Absolutely bizarre just how blithely you'll admit to regularly auto-labeling disabled people as porn

So you guys have an automated system that falsely flags marginalized people and you have moderators that make decisions that can ruin people's livelihoods in 30 seconds. Sounds like your systems kind of suck.

So let me get this straight, Aaron: you're aware your automated moderation systems are not functioning correctly, and then you want to impose much wider use of them this month?

Alt text: a drawing of a knight, heroically putting one foot in front of the other despite the many arrows that pierce him. The text, featuring an illuminated manuscript style font, says "Make it through. Make them pay"

If you automatically assume that anything any media outlet reports is a lie if a cop is the source you’re likely to be far better informed than someone who automatically assumes it’s all true.

To be fair, I am prone to blocking liberally, so inevitably sometimes they are the result of overreactions or misunderstandings on my part.