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.
I just went looking for the post on X and got a message that it didn't exist
I searched for Guri Singh and got search results showing his account existed, but when I clicked on it I got a message his account did not exist
Does anyone know if he made his account private or if he got nuked by Elon? Or do I just suck at X (because I never go there)?
I explained the concept of "blorbo from my shows" to my 71 year old immigrant grandfather because I referenced it in passing and I thought nothing of it, until today when he said "I think I'll watch peaky blinders tonight and see my blorbo from my shows" referring, of course, to Cillian Murphy playing Tommy Shelby
English isn't his first language so he's not super in touch with modern slang, so I've been accidentally teaching him to talk like a tumblr user. His favorite thing to say lately is "me when I'm a little hater" when he's like talking shit about the neighbor's son
I explained the “x before gta6” meme to my immigrant father and he, in turn, explained to me how back in his day in Romania, they had the same type of joke, except instead of it being gta6, it was about the imminent death of a singer named Gică Petrescu, who everyone was continuously shocked by because he refused to die. Every time a momentous event happened people would say, in essence: “This happened and Gică Petrescu hasn’t even died yet?!?”
So. He understood the gta6 meme immediately because they apparently had the same thing in Romania when he was young, except way, way more morbid
Every time you go in a public place and something ISN’T disgusting it’s because somebody cleaned it. Every time you feel comfortable using a public bathroom or sitting at a restaurant table or setting something on a gas station counter or playing on a playground it’s because somebody cleaned it.
Thank you to everyone who cleans the world, especially those who are underpaid and under appreciated.
[ID: a screenshot of a comic speech bubble. The black text in it reads "No matter how open-minded, socially conscious, anti-racist I think I am, I still have old learned hidden biases that I need to examine. It is my responsibility to check myself daily for my stereotypes, prejudices and, ultimately, discrimination." /ID end]
antiracism is a constant process. i was raised in a racist village and it's not easy to get rid of it. i moved away over 10 years ago but those ideas are still haunting me.
also keep in mind that shame + guilt are not conducive to growing as a person. when it comes to "checking yourself" it should be a non-judgemental process. it's not about flagellating yourself for every bad thought or trying to purify your mind of all corruption. it's only when acknowledging your own racist thoughts doesn't fill you with dread that you can really progress past the white guilt of it all.
radical self-acceptance & genuine self-critique are not opposites. they need each other. do not let obsessive-compulsive behaviors colonize your desire to grow as a person.
I work in a bookstore and let me tell you kids love brightly colored books with cartoon-like drawings of people on them! AKA alll contemporary romance books!! Many a parent gently saying “that books a little too old for you sweetie”
just so we’re clear if you’ve never actually seen a cybertruck in person and have only seen photos of them i cannot stress enough how much worse they look in real life. like i honestly don’t know how it’s possible. most things look basically the same in pictures and in real life. but as stupid and ugly as cybertrucks look in photos, every person i’ve spoken to who has seen one in real life agrees that they somehow look even worse in person. and i know you’re thinking to yourself “tah they already look so bad in photos, how can they possibly look even worse in person?” I DONT KNOW. the first time i saw one on the road i was on a phone call and i literally cut myself off in the middle of a sentence just to be like “oh my GOD.” just an incredibly, laughably, unbelievably bad vehicle. i’ve never experienced anything like it. they’re just so bad
it’s okay to have clumsy and awkward sex and it’s actually really common
the ability to laugh while you’re lying tangled and naked together bc you realize just how silly something is is really really intimate and trusting, and it’s the best feeling to be like that with someone and to be relaxed
it happens with long-term partners, it happens on hookups, and it’s rarely as bad as it’s always made out to be, except we usually think it’s bad we're told that’s not how sex is supposed to go. but that’s not true.
it can still be passionate and intimate while being messy and clumsy, absolutely. you don’t need to “perform well.” sex isn’t a show: it’s about feeling good. sometimes it takes experimenting to figure out what feels good, and fumbles and awkward moments and laughter is all just part of it. sometimes that’s a huge part of the fun.
painting this on the ceiling above my bed so it's the first thing i see upon waking in the morning and the last thing i see before falling asleep at night
murderbot and arts first meeting is literally so funny. like imagine you meet a biblically accurate angel and instead of being all 'be not afraid' it says actually you SHOULD be afraid. and then when you are in fact afraid it goes oh shit oh fuck. not THAT afraid, sorry. wanna watch tv? and then you watch tv with it and it keeps telling you to pause it when its favourite characters are in danger so it can calm down. and then it asks to do surgery on your bones. Asshole Research Transport character of all time
you have permission to pick that 2 year old "abandoned" project back up. it's not mad at you for setting it aside. and maybe time and distance have helped ease or erase the things that made you put it down in the first place.
romaincrisis
jbbartram-illu
jacobvanloon
