Some People just wanna watch the world learn. [ Jewish, Chronic Pain, Neurodivergent, White, Queer ] This blog is unapologetically pro-trans, supports inclusivity within the Queer community, and believes in uplifting the voices of minority groups to which the writer does not belong. Compassion and intersectional support is necessary to survive the world and make it better than what it is. [ Blog writer is 18+. No adult content, but most posts are untagged and I often blog about current events. DNI if this makes you uncomfortable. ]
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.
You said something in “Smith” which I hope I grasped, and there was a feeling almost of recognition. An odd feeling of grief overcame me when I read it. I cannot explain my feelings any clearer. It was like hearing a piece of music from way back, except that it was nearer poetry by Graves’ definition. Thank you very much for writing it.
Terry Pratchett, in a letter to J. R. R. Tolkien, 22 November 1967
Thank you very much for your letter. The first one that I have received with regard to Smith of Wootton Major. You evidently feel about the story very much as I do myself. I can hardly say more.
J. R. R. Tolkien, in reply to Pratchett’s letter, 24 November 1967
And he did in fact say on at least one occasion that it was this that pushed him to always engage with his own fans in the same kind and conscientious manner.
trying to make a creative project without men and uproot any masculine words will drive one insane. There's the obvious stuff, right, swordsman becomes swordsmaid, count becomes countess, gladiator to gladiatrix and so on.
But did you know that "-er" is a masculine suffix, for which "-ster" as in "sister" is the feminine equivalent? Baker means a man that bakes, the historical feminine equivalent is a bakster, a webster is a female weaver. Some words already have the feminine as default, youngster, teamster, mobster, but most? Trying to be a principled feminist has you saying shit like goonster
for real! Lots of other examples too, anything "patr" comes from father, "patriarch" is the obvious one, but patriot, patron, patronym all would need "matr" for mother is instead. Pope and Papacy derive from "Papa" and would be Mome/Mamacy if you want yuri catholicism (hellworld!!) it's EVERYWHERE
My favorite example, beyond the obvious ones like lord/lady, waiter/waitress, steward/stewardess, is housekeeper, which is now the more acceptable term for a maid, when maid, of course, just means "woman" and keeper is masculine, would be keepster or keepess but (to my knowledge) neither were used historically because, well, keeping wasn't something women got to do!
Likewise, waitress has fallen out of favor for host or server, which are also both gendered terms. (hostess or serving girl, respectively. Servant is also gendered, maidservant and servantess were used)
There's been a concerted push with a lot of these to phase out the feminine variants of these words and just treat the masculine variants as neutral, or find neutral alternatives like fireman -> firefighter (even though fighter is also gendered) because the feminine variants are seen as inherently lesser, which, yeah that's how women are seen.
There is a line you have to draw somewhere of "okay let's just treat this and below as gender neutral because, frankly, most words that just mean "person" mean "man" historically because women weren't (and still largely aren't) considered people, and otherwise we're going to shred the whole English language" and I get that.
BUT. I think a lot of folks draw that line at the start and insist that man/guy/dude/bro can be gender-neutral which is obvious stupid. And I think it's always worth having this investigation and questioning how we speak, it shapes how we think about worlds and people real and fictional.
Okay one last silly aside about barista, which is supposed to be gender-neutral from Italian but men got weird about it and then invented baristo so now -ista is kinda feminine? Typically it's just borrowed into English to sound Foreign (Sandinistas leading to the exonym Corbynistas)
Bucket, Captain Lieutenant are gendered masculine btw, it'd be buckette captaine and lieutenante. It's not just an issue of suffixes being "ignored"(masculinized) when borrowed from french because we write Debutante with the e. Multiple times a month I stumble into new examples, the battle never ends
hey i'd just like you to know ever since i saw this post it's been the only thing i think about when at work. Because of it i've tried imagining a version of english that is more feminine or even 100% fem.
1. it's been such a wild ride just going "my god how we speak is weird."
2. I has been a pure joy coming up with an idea for a world for a bee species and a lesbian lizard species to speak in this language.
I started thinking about what this implied about "gangster" and immediately envisioned a gang of women dressed in natty 1920s-vintage suits and carrying tommy guns and is it hot in here or is it just me?
Ultimately, she spent 20 hours redoing the copy from scratch — and with her $100-per-hour rate, that meant her client was shelling out $2,000 for copy that likely would have ended up being far cheaper had a human just written it in the first place.
I'm an electrical engineer and for the longest time I was saying that electricity and electronics isn't magic, but think about it.
You literally have to collect rare stones from remote locations, put them into specific formations to work. All of this gets written down in symbols which don't make sense to the uninformed. It gets powered by energy which can not be seen in most cases.
Like what else do you want. What's your standard for calling something magic.
i do actually get tired sometimes of having the same basic decency discussions around native culture. stop using the term spirit animal. don't use the skin word for shapeshifters. i saw a plushie kickstarter with the W word for a small skull deer plushie and it had already met it's goal. people just blithely repeating terms with no connection to their culture. actively profiting off of it.
yes! though "offensive" is a bit reductive; it's not a swear or a slur, it's just a concept that the navajo would prefer non-navajo not to talk about, both because of their belief about how using it's name invokes/summons the being in some way and because of the ongoing appropriation and erasure of native culture.
so apparently netflix is coming out with another season of the tv series "Skin--- Ranch".
reminder that the original name of the ranch was Sherman Ranch and you can refer to it as such. secondary reminder this is all being driven by the white mormon ranch owner as a PR piece for his shitass UFO ranch.
Modern research shows the public work together selflessly in an emergency, motivated by a strong impulse to help
“The notion that people panic and run screaming for the exits is a Hollywood fiction,” said Prof Stephen Reicher, an expert in group behaviour at the University of St Andrews.
“Characteristically, people stay and help each other,” he said. “We found this during the 7/7 attacks on the underground and the 1999 attack on the Admiral Duncan pub in London, where people looked after each other even though they feared other bombs.
“In our own research on the Leytonstone tube attack in 2015, there was an amazing level of spontaneous coordination by bystanders: some directed others away from danger. Some distracted the attacker. Some confronted the attacker. Each was able to act because of the others. Heroism was a feature of the group, not just the individual,” he added.
Prof Clifford Stott, a specialist in the psychology of crowds and group identity at Keele University, agreed. Modern research, he said, showed “bystander apathy” was a myth. Instead, strangers often work together in emergency situations with highly sophisticated unity.”
Bystander apathy is a myth invented by the New York Times to cover up that the police were called by several residents of the building, but the cops refused to act. The cops then told the Times that 38 people just watched her die (a seemingly arbitrary number and a physical impossibility based on where the attacks occurred), and the Times ran with it. In fact, Kitty was alive when the cops got there, and was being held and comforted by one of her friends who lived in the building because one of the people who saw her get attacked from across the street called her friend to go get her. Because people care.
You have just been attacked. How likely is it that someone will come to your help? If you remember the infamous case of Kitty Genovese in 19
I will always re-blog this. The story of Kitty Genovese’s murder has gone down in history as a story about everyone watching it happen and doing nothing and none of the story is true.
janainaart
snowghoul
origamiaround
yousharknotpass
annaxmalina
jacobvanloon
dailybridgerton
germansingergirl
