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.
The reason why so many of y'all's feminism sucks is because you still believe deep down in your hearts that there are only two kinds of people in the world: precious, ethereal, fragile dollthings called "women", and violent, lustful, rage-fueled apes called "men". Until you throw that idea away, 3rd-grade-tier "girls rule boys drool, girls are princesses and boys are stinky :(" is as feminist as we'll ever get-- and I hope it's obvious that that's lightyears away from the bare minimum of where we need to be.
I don't know how I'm supposed to explain to ostensibly trans-friendly feminists that "women are beautiful soft things made of glass, men are obsessed with violence and sex" is exactly what the patriarchy wants you to believe. Patriarchy wants you to believe that being a woman and/or having a vagina (patriarchy generally believes those two things are synonymous) makes one shatter on impact with reality. It makes you easier to control if you are scared shitless of the other half of the population, and it makes you more compliant with your lot in life if you believe it is in the nature of the other half of the population to rape and kill rather than realise those were choices those individual rapists and murderers made. There is no way to make gender essentialism progressive and feminist, because it is one of patriarchy's tools of subjugation. Stop trying to make it progressive.
And I can scream all of that from the rooftops over and over again, and what I hear in reply is "Trans men really are men because no woman would ever decide to become an inherently evil repugnant rapist ape", and "You're so right. Trans women are women because they too are pretty delicate little objects I can fuck", and "You're non-binary? So are you fucktoy non-binary or sexpest non-binary?", and my patience runs ever thinner.
by the way, if you ever want to actually be on a jury and actually participate in the judicial system, you need to not say this kind of shit during the selection process. they will not select you. you will not get to convict. you need to be the most wishy-washy, "I don't know, I think there's many sides" austerely "I believe in the letter of the law" type motherfucker to get selected. you have to come off as someone who is profoundly disconnected from politics etc. You can't lie, obviously; that's perjury. But you can be kinda vague and noncommittal and that's not a crime.
Most especially, if you feel the temptation to "expand" on your views, as this prospective juror put it, DON'T. Do not discuss anything but the direct answer to the questions - don't discuss your feelings, motivations, or do any other pontificating unless you are directly asked, even if they use long uncomfortable silences to make it seem like you have to keep talking. If they ask you a yes or no question, answer "yes," "no," or maybe "I don't know" or "it depends," full stop, without elaborating unless they ask.
Give this a read if you'd like to be a jury to your peers some time:
A resource to help prospective jurors prepare before going to the courthouse to answer questions truthfully but neutrally so as not to be ex
something i've noticed that has become really annoying in the past 10 years or so is this fad of what i've been calling, for lack of a better word, "structural whataboutism." it's that thing where, when faced with a concrete, resolvable problem in your community, your answer is to blame it on a vast, unsolvable issue of structural inequality and then throw up your hands. "there's trash all over the ground in this corner of the park" becomes "well, that's where MEN OF COLOR congregate after their 12-HOUR GRAVEYARD SHIFTS and i'm not going to support a CARCERAL SOLUTION to a CAPITALISTIC PROBLEM. WE NEED TO ELIMINATE POVERTY AND THE SUBJUGATION OF THE WORKING CLASS" and it's like okay but sis. someone still has to go pick up the trash. we don't need a carceral solution, we need more trash cans. you're not going to eliminate poverty and the subjugation of the working class and even if ya did, there would still be trash on the ground. how any of this passes for radicalism within their peer groups i simply don't understand. it's radical laziness more than anything else
I was on a canoe trip once with a river biologist who worked for the county. After we found and removed a car tire, she started talking about the annual river cleanup her department organized. From a water quality or ecological standpoint, removing shopping carts, car tires, and other macro trash from the river really wasn't that important, she said. The real threat to the river was industrial and agricultural runoff.
"But!" she said:
People who see a clean, trash-free river are more likely support laws to curb more harmful "systemic" forms of pollution. People who participate in river cleanups take pride in their work--their river!--and become evangelists for protecting it.
Immediate action leads to systemic awareness, which leads to systemic change.
Saying "there's no point in doing something small until the big thing is fixed" is literally just the Glorious Revolution Rapture story all over again, and it's not helpful.
Not gonna lie, I'm a little annoyed that despite the vast amount of project hail Mary spoilers I saw on tumblr in the last few weeks (and yes I know that not blocking the tag was my own fault) I did not see a single post about the huge flashing light scene in the middle of the film. Hell, it was set right in the middle of a very important plot scene. But the flashes were so bright and so intense that I had to close my eyes and cover them with my hands for I think nearly two minutes. And I'm not even epileptic! I just get migraines. Thankfully I went to see it with a friend who could tell me what was happening in the scene.
So, I'm going to warn people who may not have seen the movie yet. When the ship is pulling away from the green planet, the music will calm down, then a bang will be heard. That's when the flashing lights start. They'll pretty pretty much keep going until there's a scene change and the audio cuts out. That's when you can open your eyes.
The European Union already forced Apple to abandon its proprietary charging port and adopt USB-C across its entire iPhone lineup. It just did something bigger. A new EU mandate requires every smartphone sold in Europe including Apple devices to feature a battery that can be replaced by the user without specialist tools, without voiding a warranty, and without sending the device to a manufacturer approved service center. Batteries must maintain a minimum capacity threshold after a set number of charge cycles and replacement parts must remain available for up to ten years after a model goes on sale.
The consumer electronics industry built its current business model around batteries that degrade, cannot be replaced at home, and create a natural upgrade cycle every two to three years. The EU just legislated that model out of existence in the world's largest regulatory market.
Apple, Samsung, and every other manufacturer now faces a choice between redesigning their devices for the European market or accepting that their current hardware architecture is no longer legally sellable there.
Given that no company walks away from European consumers voluntarily the phones are going to change and once they change for Europe the rest of the world will ask why theirs still do not.
Okay I have things I should be seeing to but I couldn't help myself. In case you, like me, have not read all of these stories and would like to be amongst the lucky 10,000 today:
I Have No Mouth and I Must Scream by Harlan Ellison
The King in Yellow by Robert W Chambers*
The Lottery by Shirley Jackson**
The Masque of the Red Death by Edgar Allan Poe
The Monkey's Paw by W.W. Jacobs
The Most Dangerous Game by Richard O'Connell
The Nameless City by HP Lovecraft
The Ones Who Walk Away From Omelas by Ursula K LeGuin
There Will Come Soft Rains by Ray Bradbury
The Yellow Wallpaper by Charlotte Perkins Gilman
The Veldt by Ray Bradbury
Honorable Mention from the comments/reblogs:
All Summer in a Day by Ray Bradbury
*note: this is actually a collection of short stories and clocks in at about 72k words
**Originally published in the New Yorker in 1948; interestingly, the New Yorker still has this story archived on their website BEHIND A PAYWALL. CAN YOU IMAGINE.
All the Troubles of the World by Isaac Asimov. This one made me lie awake in bed all night for like a week after hearing it - our 8th grade teacher read it out loud to us one day in class. At the end of the year before we 'graduated' I remember going back and asking her. I still have her note in my scrapbook with the name of the story so I could look it up on my own (along with some other book recommendations and the book she had been reading to us but hadn't finished - The Wind Singer.
Pigeon steals poppies from the Tomb of the Unknown Soldier, Australian War Memorial, Canberra, Australia in order to build a nest beside a stained glass window.
This is two pigeons, pigeons nest in bonded pairs (notice the first one is checked and its mate on the nest is barred). Usually they don’t make nests nearly so big but I guess if you have the materials, go for it.
The nest is so unusually big because the vast majority is a platform to keep the actual nest (just that tiny ring in the corner around the bird sitting in it) cushioned from the anti bird spikes.
This is a work of beautiful defiance.
Using the very thing installed to make just a moment’s rest impossible as structural supports for an immovably stable nursery.
The symbolism achieved by these pigeons is better than some writers can hope for and I love it!
From the nest on the bird repellent spikes to the fact that those spikes are along the stained glass windows of a church, a place associated with sanctuary and compassion. The fact that the nest is made of stolen poppies for remembrance day hits the hardest though. Of the 54 animals to be awarded the Dickin Medal for acts of gallantry during WW2, 32 of them were pigeons. These were messengers who flew through battlefields and across borders, many of whom were killed or severely injured by enemy forces including gunfire and trained falcons. Many of their achievements saved the lives of hundreds of soldiers, and yet now their descendants are faces with anti-bird spikes, shooting and poisoning in an attempt to rid the cities from the rats with wings. I love this picture because it feels like they’re taking back just a little bit of that credit owed to them.
annaxmalina
blakekathryn
izzy806
janainaart
origamiaround
tangentwithtanner
theartofmadeline
