Text of tweet under the cut because it is loooong.
But... Stochastic Parrots.
Timnit Gebru was fired from Google in December 2020 for refusing to retract a research paper, and every single warning that paper made about large language models has now happened at a scale the industry spent 4 years trying to make people forget about.
Her name is Timnit Gebru.
She co-led the Ethical AI team at Google. She co-wrote a paper called "On the Dangers of Stochastic Parrots" with Emily Bender at the University of Washington and two other researchers. The paper was 14 pages long. It was submitted to a top AI ethics conference. And it was the reason Google decided that one of the most senior Black women in AI research could no longer work there.
The story Google told publicly was that she resigned. The story she told, confirmed by 2,695 of her colleagues in an open letter, was that she was fired by email while on vacation because she refused to either retract the paper or remove her name from it.
The paper had not even been published yet.
Here is what she actually wrote, and why every prediction inside it has now come true.
The first warning was about scale itself. Bender and Gebru argued that training ever-larger models on ever-larger scrapes of the internet would produce systems that appeared fluent but had no actual understanding of language. They called these systems stochastic parrots because they would repeat patterns from training data with statistical confidence and zero comprehension. The paper predicted that this apparent intelligence would fool both users and developers into trusting outputs that were structurally incapable of being reliable.
This was 2020. GPT-3 had just come out. The paper predicted the hallucination problem before anyone had a word for it.
The second warning was about bias amplification. The paper documented in detail that internet-scale training data contains systematic overrepresentation of dominant viewpoints and underrepresentation of marginalized ones. The models would not just absorb this bias. They would amplify it, because the optimization process rewards confident outputs, and confidence in language patterns tracks frequency in the training set.
The prediction was that hiring tools built on these models would discriminate against women. That healthcare triage tools would underperform on Black patients. That loan approval systems would entrench inequality while presenting their decisions as neutral algorithmic judgment.
Every one of those things has now been documented in deployment.
Amazon's hiring algorithm penalized resumes that contained the word "women" in any context. Healthcare risk scoring algorithms used by major US hospitals were found to systematically underestimate the medical needs of Black patients. Apple Card's credit algorithm gave wives credit lines 10x lower than their husbands for the same financial profile.
The third warning was about environmental cost. The paper calculated that training a single large language model produced emissions equivalent to the lifetime output of 5 cars. The prediction was that the race to scale would create an environmental footprint that would eventually rival entire industries.
In 2024, Google's emissions were up 48% from 2019, and the company explicitly blamed AI infrastructure. Microsoft's were up 29%, same reason. Both companies have now quietly abandoned the climate commitments they were publicly celebrating the year Gebru was fired.
The fourth warning was about documentation. The paper argued that the training datasets being assembled were too large for anyone to actually audit. Nobody at Google, OpenAI, Meta, or any other lab could tell you with confidence what was in the data their models were trained on. This was not a temporary problem to be solved later. It was a permanent feature of the approach.
In 2023, researchers discovered that the LAION-5B dataset, used to train Stable Diffusion and other major image models, contained thousands of images of child sexual abuse material. The companies that had trained on the dataset had no way of knowing. The paper predicted that category of failure 3 years before it was found.
The fifth warning was the one Google cared about most.
Bender and Gebru argued that the deployment of these systems would centralize linguistic and cultural power in the hands of the small number of companies that could afford to train them. The internet would become a place where the dominant voice was a statistical average of dominant voices, presented as a neutral assistant. Languages underrepresented in the training data would degrade over time as more web content was generated by these systems and fed back into the next training run.
This is now happening in real time. A 2024 study found that 57% of new web content in English is AI-generated or AI-assisted. Researchers studying low-resource languages have documented active degradation in translation quality, because the synthetic content fed back into training is itself worse in those languages.
The paper Google fired her for predicted the model collapse problem before model collapse had a name.
The mechanism behind why this all happened is the part of her work that nobody quotes.
Gebru's argument was not that AI is dangerous in some abstract sci-fi sense. Her argument was that AI is dangerous in a very specific structural sense. The technology was being built by a small group of researchers who shared similar backgrounds, worked at similar companies, and were rewarded for shipping products faster than competitors. The incentive structure made it impossible for safety, ethics, and bias concerns to slow anything down. Anyone inside the system who raised those concerns was either ignored, sidelined, or removed.
She was making that argument from inside Google.
Then Google proved her right by removing her.
The team Google had built to make sure their AI was safe was dismantled in 90 days because they did the job they had been hired to do. Margaret Mitchell, the other co-lead of the Ethical AI team, was fired two months after Gebru for searching through her own emails for evidence of how Gebru had been treated.
Gebru did not stop. She founded DAIR, the Distributed AI Research Institute, in 2021. The mission is to do AI research outside the control of the companies that have a financial interest in not hearing the answers.
Every prediction in the Stochastic Parrots paper has now been validated by deployment. Hallucinations are an industry-wide problem the largest labs cannot solve. Bias amplification has been documented in hiring, healthcare, lending, and criminal justice. Environmental costs are larger than entire small countries. Training data audits remain impossible. Model collapse is an active research crisis at every major lab.
The question worth sitting with is the one almost no one in the industry will say out loud.
Every researcher with the technical credibility to call out these problems watched what happened to her in December 2020 and made a calculation about their own career. The number of people willing to speak publicly about safety and ethics issues inside the major AI labs collapsed after that firing and has not recovered.
The researcher Google fired for warning about exactly what is now happening was right.
The company that fired her is now the second-largest deployer of the technology she warned about.
And the people inside that company who agree with her are not allowed to say so.
as a kid I hated Dr Seuss illustrations because no one ever acknowledged that they were creepy. I think if anyone had ever indicated that I was supposed to find them a little unsettling I would have been really into them. but instead I felt gaslit and that really upset me.
also teachers and librarians and stuff would invoke the cat in the hat like he was a friendly face I should be happy to see and not an extremely sinister entity I would not want to come to my house. hello. hello can anyone hear me.
"That ...thing... is no cat! It is an eldritch abomination, here to sow chaos and destruction! It should not be here when your mother is out! It should NOT BE HERE! GET IT OUT!!! GET IT OUUUTTTT!!!" -your pet fish, probably
House of Leaves is a book about a man trying to explore the inside of his new home, which proves substantially more difficult than should be possible, due to its non-Euclidean dimensions.
Except that House of Leaves is actually about a documentary about a man trying to explore the inside of his new home, which proves impossible to document properly, due to its non-Euclidean dimensions.
Except that House of Leaves is actually about an academic paper, which itself is about an alleged documentary about an alleged non-Euclidean home, but none of this can be verified.
Except that House of Leaves is just, like, this fucking stack of papers I found in this old dead guy’s apartment, and, like, I don’t even know if its his or what because like both his eyes were gone and he didn’t have any degrees or anything. I tried asking my friend from the tattoo shop about it but she was fixing up this thumper tattoo right above a girl’s pussy the whole time we were talking; like Thumper from Bambi, and I’m not gonna lie man I was pretty distracted and I just couldn’t [XXXXS XXXX XXXX. XXXX XXXX XXX X XXX XX XXXXXXX XXXXXX XX XXXX XX XXXXXX XXXX XXX]^*
———
* Editor’s Note: House of Leaves is a metatextual examination of the ability for a minor inconsistency in measurement, which might have been ignored by a normal person, to cause an all-consuming obsession spiral in the sort of person who is academically trained, obsessed with documentation, or overtly pedantic, and-
“Johnny, please, I’m your Mother; I know I have a lot to apologize for but I can’t, not unless you return my letters; are you even getting my letters? I send one every week you know, but I think that nurse keeps them from you, keeps you from getting my letters, because I know, I know my sweet Johnny would never leave his poor old mother rotting away by herself in this horrible drafty ward if he was getting his mother’s letters, would you, sweet boy? I know she’s stealing my letters from you, she’s always touching my things, even though they haven’t moved I know she’s touching them I know she keeps you from me I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know [there’s a limit to what you can know. Some folks bump up against that limit and bounce right off and keep swimming, and some folks, well. Some folks SPLAT]I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know [they just gotta keep digging away at it. Even when it’s not that deep, they just… grab a shovel and keep digging. Lord knows what motivates them. A kind of Madness I suppose. Some folks find GOD in that madness. Not much of a believer myself]I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know I know
I know
I know
I know
I know
I know
I know you’ll visit me for Mother’s Day, won’t you?
I'm not up to date on the City Council of Darkness right now so maybe they do clarify this but I'm a bit confused as to how exactly the coterie backing their own dog mayor would help them. I mean the point is that the dug mayor can't vote, which gives the two rich families the majority vote between them. Bingo also would not be able to vote. He is a dog. The balance of power stays the same only someone else gets Instagram ad money.
it has been clarified that they plan to teach the dog mayor to vote. They will make a button board or something with the various votes on it, then train the dog to hit the desired vote button at a signal from one of the vampires.
everyone's talking about the ibs/autism haha funny comparison thing while I'm still stuck on the concept that hamsters exist in the wild. like naturally
tf do you mean they're a wild creature. you find those ankle-biters at the pet store
Not a biologist, but the evolution of mammals is way more granular than you might expect. Humans are the sole surviving species of the genus Homo, which was a real party before the other ones went extinct. You're in for a fun time.
Domestic cats are believed to be domesticated not from tigers, but from the African wildcat:
Which evolved to be small just because it's sometimes more useful to be small.
And no, hamsters are not off-brand rats. They're part of the rodent order, which includes beavers, moles, capybaras, guinea pigs (yes, also wild) and lots of other fun things:
I think the most charitable interpretation of this is that people connect to characters that are like them, and that the experiences of Blackness is so alien to them that the White characters are the only ones that White people connect with. Combine that with the fact that Tumblr is so White as a platform, and you get this issue.
It’s still bad and racist, but it is understandable. People should still do better though. (I’m not trying to justify, I’m trying to come up with an explanation)
"So my charitable interpretation is that they're horribly racist" they are but the fact that you think that's a small enough deal to call it a charitable take is concerning.
"Tumblr is so White as a platform" and why do you suppose that is? did this Whiteness spontaneously generate in a vacuum? or could it be that black bloggers were repeatedly harassed and kicked from the platform over and over again?
There are a lot of things I'm sad about in my life. You don't get to go through the kind of medical trauma I've been through and come out unscathed on the other side.
But one thing I'm really bitter about is that I can't remember my wedding anymore. The pernicious anemia took it from me and wiped my brain clean. Except it's not clean, not really. I remember it in patches. Like red wine stains on a white rug that have never quite lifted out no matter how hard you try.
I look at the pictures on my bookcase, and they feel like remembering a story someone else has told me. There's a young woman in a white dress wearing my face, and she looks happy. I'm happy for her. But you can see the strain around her eyes, too. The pain she's hiding because no one with authority believes her when she says her body doesn't feel right. That something is Wrong.
They won't believe her for another decade. They won't believe her until it's almost too late, and it's that lateness that will rob her of her memories and turn them into a wavering rainbow suspended in the fine haze of watery sunlight that occasionally surfaces through the blanks.
There's one memory that's real, though. Solid. It's not my vows. It's not my father walking me down the aisle. (Though those are there, just hazy and dream-like). It's our first dance.
It's the lights dimming around the room as the staff cleared the floor, causing the fishbowls full of white roses and LED lights on the tables to wobble like pools of moonlight against dark paneled walls.
It's the band inviting us out onto the floor and us giggling because we know what's coming next, and no one else does. It's the twang of a banjo reverberating around the room through the speakers, followed by the dulcet tones of Kermit the Frog wondering why there are so many songs about rainbows.
It's us waltzing around the enclosed circle of light, singing to each other out of tune and grinning like idiots as everyone around us starts to laugh.
It's everyone joining in on the song because it's the Muppets, and everyone knows the words. It's 100+ people singing the Rainbow Connection, some laughing, some a bit tearful, because it's bringing back memories. Because it's making a new one.
It's looking up at my new husband through the brain fog and all the pain in my body and thinking, "I want to remember this moment forever."
I don't know what entity was out there listening to me at that moment and chose to grant that wish. I don't know why this is the one memory that stuck while everything else in my brain got decimated into scattered, fragmented snapshots. But I'm so, so thankful it is.
romaincrisis
stormborns
baratheonpilled
snowghoul
2087
jbbartram-illu
titsay
annaxmalina
