
#extradirty

if i look back, i am lost
Misplaced Lens Cap

oozey mess
DEAR READER
we're not kids anymore.
Xuebing Du
Sweet Seals For You, Always

blake kathryn
Peter Solarz
TVSTRANGERTHINGS
Monterey Bay Aquarium
art blog(derogatory)
NASA

roma★
KIROKAZE

No title available
Cosmic Funnies
trying on a metaphor

Kiana Khansmith
seen from Sweden

seen from United States
seen from Canada

seen from South Africa

seen from Germany
seen from United States
seen from United States

seen from United Arab Emirates

seen from Paraguay
seen from United Arab Emirates
seen from United States
seen from United States

seen from United Kingdom
seen from United States
seen from United States
seen from United States
seen from United States
seen from United States
seen from United States

seen from China
@dittorom
[source]
🚨BREAKING: OpenAI published a paper proving that ChatGPT will always make things up.
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.
Every time ChatGPT gives you an answer, ask yourself: is this real, or is it just a confident guess?
Let me save anyone who's curious a step: Here's the paper on ResearchGate
Stop me if you hate the concept: short, fat, hairy lady gets isakai'd into a high fantasy, and instead of "oh look at all these ethereal elves woe for I am but a flawed mortal" routine she lands in Dwarf territory and is immediately revered as the most enchanting and desirable maid in all the land. This immediately becomes a zesty romantic drama. Thoughts
Frankenstein (2025) + trivia
RED WINE SUPERNOVA 'Behind The Curtain' (2023) Chappell Roan, dir. Ryan Clemens & Hope Alexander
I love that every once in a while brushbug gets a sudden burst of notes again. We all love brushbug.
Stay safe, guys
Get immediate help 24/7/365. The Trevor Project's crisis counselors are just a text, chat, or phone call away. Completely confidential and f
it’s three taps anywhere on the screen on a smartphone!
This is important punks. Deadly important.
I’ve texted their hotline before. It was super helpful and even if it hadn’t been the amount of time you’re there can be enough to let your urges fade and stay safe.
Leaving this here:
The Money Tubbs only comes around every 5628 seconds. Reblog the Money Tubbs and you’ll find money!
Bitttchhh the last time I reblogged some bullshit like this I booked a 2k 30minute shoot lmao
I received 2k 2 days after reblogging this
where is my beloved "its ween" pumpkin. i require its presence
Pretty sure this is it?
YES MY PRECIOUS
ITS FUCKING WEENNNNNNN
why is this post breaking containment. don’t do this to me
IT'S WEEN
theo van gogh was the one who suggested that his older brother vincent start seriously painting. as soon as theo was gainfully employed he gave vincent around 15% of his own yearly salary for art supplies, lodging, and food. about 2/3rds of vincent's surviving letters were to theo (including vincent's earliest and last letters), all of which were found stored in theo's desk. theo's child, vincent willem, was born on january 31st, 1890, and vincent was so delighted by his nephew that he painted almond blossoms for him. vincent shot himself half a year later on july 29th, 1890. theo's distress at his brother's death worsened his syphilis symptoms and he died half a year after his brother on january 25th, 1891 (four days before vincent willem's first birthday). theo was reburied next to vincent in auvers-sur-oise at the request of theo's wife johanna.
Almond Blossoms, 1890, Vincent van Gogh
And that love lived on Theo's wife, Johanna, who was the one who pushed for the preservation of Vincent's paintings. Johanna who made sure that her husband's beloved brother would not be forgotten. Johanna who fought tooth and nail so that Theo and Vincent would never be forgotten. Johanna who carried the family legacy, who made sure that the works of Vincent would be kept in her possession. And then Vincent Willem, named for a uncle so loved, carried this legacy and founded the Van Gogh museum in Amsterdam.
The van Gogh family history is deeply interwoven with love. It was the love of a brother that gave a young man the courage to paint and the resources to do so. It was the love of a woman for her husband and her husband's beloved brother that made that story known. And it was the love of a nephew, who was so deeply loved even if for such a short time, that made it possible for the world to know Vincent van Gogh.
it’s almost as if Tolkien knew what he was talking about
It’s almost as if Tolkien personally witnessed one of the bloodiest wars in human history and then used that to fuel his writing.
This is the end result of anti-trans propaganda: more harassment of girls and women, more policing of gender roles, more gender McCarthyism.
FFS, THIS IS SHIT I FOUGHT IN THE 80s.
transphobia isn't just an attack on trans people, it's about enforcing rigid and patriarchal rules on womanhood, femininity, and beauty (read; the white standard).
Anyone who doesn't strictly conform will be punished.
My cis daughter got physically blocked from using the girl's bathroom for wearing a dinosaur shirt to gymnastics. She was seven. Gender policing is a plague and a blight.
I had a haircut exactly like this in 1986-87, starting when I was 9. People were far less fucking weird about it then, which is depressing to consider.
saw some debate over the english title translation of dungeon meshi and I am here to offer you this alternative
DUE TO POPULAR DEMAND LAIOS FIERI IS NOW A PRINT:
🔗 🖼️ https://www.inprnt.com/gallery/steohsama/dungeons-drive-ins-dives/
This is a gallery-quality giclée art print on 100% cotton rag archival paper, printed with archival inks.
i swear to god if one more stupid fandom ruins a beautiful text post i am calling the police
I am obligated to reblog this again, because it is now Superwholock, and therefore perfection.
people need to remember that every tumblr post in 2012 was like this
Keep up the great work, lads
I love how this addition would be absolutely incomprehensible almost anywhere else but we know exactly what it means
I can hear the cartoon bonk and coconut noises resonating from this 🤣😭