DeutschlandDenDemokraten

We all keep joking the tumblr fandom would fit on a bus

would it tho???

i wanna keep headcount of all the pigeons!

Please reblog from only ONE of your blogs!

My favorites:

73% of people globally feel safe walking alone at night, the highest ever recorded

UN high seas treaty ratified, allowing for nations to extend major protections to marine environments

Extreme poverty falling

China, once one of the major emitters, is investing heavily in green and renewable energy

India's tiger populations has doubled in the last decade

Amazon deforestation in Brazil is at its lowest level in 11 years

My favorites:

73% of people globally feel safe walking alone at night, the highest ever recorded

UN high seas treaty ratified, allowing for nations to extend major protections to marine environments

Extreme poverty falling

China, once one of the major emitters, is investing heavily in green and renewable energy

India's tiger populations has doubled in the last decade

Amazon deforestation in Brazil is at its lowest level in 11 years

Not sometimes. Not until the next update. Always. They proved it with math.

Even with perfect training data and unlimited computing power, AI models will still confidently tell you things that are completely false. This isn't a bug they're working on. It's baked into how these systems work at a fundamental level.

And their own numbers are brutal. OpenAI's o1 reasoning model hallucinates 16% of the time. Their newer o3 model? 33%. Their newest o4-mini? 48%. Nearly half of what their most recent model tells you could be fabricated. The "smarter" models are actually getting worse at telling the truth.

Here's why it can't be fixed. Language models work by predicting the next word based on probability. When they hit something uncertain, they don't pause. They don't flag it. They guess. And they guess with complete confidence, because that's exactly what they were trained to do.

The researchers looked at the 10 biggest AI benchmarks used to measure how good these models are. 9 out of 10 give the same score for saying "I don't know" as for giving a completely wrong answer: zero points. The entire testing system literally punishes honesty and rewards guessing.

So the AI learned the optimal strategy: always guess. Never admit uncertainty. Sound confident even when you're making it up.

OpenAI's proposed fix? Have ChatGPT say "I don't know" when it's unsure. Their own math shows this would mean roughly 30% of your questions get no answer. Imagine asking ChatGPT something three times out of ten and getting "I'm not confident enough to respond." Users would leave overnight. So the fix exists, but it would kill the product.

This isn't just OpenAI's problem. DeepMind and Tsinghua University independently reached the same conclusion. Three of the world's top AI labs, working separately, all agree: this is permanent.

i can recognize white people as a whole have systemically oppressed POC in america, which is where i live 

most people when they make white people jokes only mean the shitty white people and i am not a shitty white person 

And hey, you know, the same worry that teen girls might regret these things is true for adult women, too. Maybe we should make them wait until they're married, in case their future husband disapproves. Maybe we should make them wait until they've had kids - what if she regrets it and wants children? what if she regrets it and wants to breast feed? what if her stupid little woman brain hasn't thought this through and correctly pinned her entire self-worth and value as a person on her ability to act as a sex object and baby incubator? what then???

Trans men aren't women, and infantilizing them as poor helpless girls is obviously transphobic. But in addition to that, even if we were to buy into the central transphobic premise here, the conclusions would still be sexist as hell. Let's say that there are no trans men. Let's pretend for the sake of argument that trans men simply don't exist. Infantilizing teen girls out of being able to make their own medical decisions isn't feminist either.

eres como un angel de la guarda

Looking it up now... What the fuck??

Description posted online:

VTuber Mimi Yanagi Arrested in the United Kingdom Over Her NSFW Commission In a surprising turn of events, British police have arrested popular VTuber and illustrator Mimi Yanagi simply for possessing her own artwork. Mimi recently shared that she was taken into custody on April 20. Although she is now out on bail, the authorities seized all of her electronics. She is also dealing with some incredibly strict bail conditions. For example, she is currently banned from posting any adult content and is only allowed to operate under a name that the police have explicitly approved. This case sheds light on the incredibly strict laws in the UK regarding fictional artwork. Even if the illustrations are entirely imaginary, British law classifies certain types of drawings as restricted material and treats them on par with illegal child pornography.¹ This strict legislation is exactly why the police were able to confiscate all of her computers and equipment.

Haven't verified this yet, this is all very new info. Very concerning to hear though, of course.

The evidence is overwhelming that there is just no scientific basis for this shit. They might as well have cops do palm-reading on suspects.

The fact that so many Americans believe that these things actually work is proof of just how unquestioning and susceptible people can be to established "truths."

So, for starters, I didn't know anything about these recent events cuz I'm an idiot 😅

But hey screw it, I'm here for a good time, not an accurate time:

This is a surprisingly complicated issue, and it's closer to actual illegal censorship than you might suspect.

That's because in the US companies are NOT allowed to OFFICIALLY discriminate based on gender, sexual orientation, or race. But you can probably already see the problem: as long as your don't have an official rule that explicitly targets a protect groups, it's incredibly hard to prove discrimination. Because there's two distinct possibilities:

Historically, this sort of selective enforcement has been a convenient workaround that allows companies to continue de-facto discriminating against protected groups they don't want around.

However, it's probably much more common for this sort of thing to happen due to unconscious biases.

In other words, people will see a photo of a transgender woman, and they legitimately, sincerely, will think it is overly sexual and breaks the rules - even if a similar photo of a cis woman would raise no alarms.

Now, there are many potential solutions that companies can take to combat this:

More transparency for moderation decisions - like, I don't know, maybe explaining who the fuck is doing the moderating and how? I mean however stupid Reddit might be with moderation, you have to admit that it lets you see exactly who is doing the moderating and you can contact them directly. Nothing like that exists here.

More diverse people doing the moderating

Training to explain what to watch out for, so that people can more easily notice when they might be experiencing unconscious bias

Having a neutral, outside organization that can provide some oversight is probably the gold-standard, in my opinion. They could audit practices and look at samples of who have been removed and why.

This is going to shock you, but Tumblr is just not going to want to do any of those things on their own. They assume that they can do whatever they want and people will just put up with it.

Interesting theory on their part, considering how unsuccessful this site is doing overall. Interesting theory considering how crucial queer people and a variety of minority groups are to propping up this failing site. Imagine what the site's gonna be like once they leave:

Users can apply their own pressure to the company in hopes of changing policies, but I'm not very optimistic about that. After all, Tumblr has been banning vast swaths of people with little explanation for years and years, and stunningly the site has been having less and less users. WHO COULD HAVE GUESSED THAT!?

My guess is that the swarm of feral weasels responsible for running Tumblr will postpone any sort of changes until this site has lost enough users to be past the point of no return. It's a shame, but hey, if they want to destroy their company, capitalism gives them that right I guess?

a world without trans people has never existed and never will