hey boss i can't come in today it's a sunny day and there's a lovely breeze coming in through my window, yeah it's rustling the branches of the tree outside that's finally bloomed so it's pretty serious
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.
If you use Firefox, you can go to the about:config page, search for "media.mediasource.enabled" and double click on it to set it to false. After you restart Firefox, all youtube videos will load entirely even when paused! This also affects other streaming websites :)
go to About:config
find media.mediasource.enabled and toggle it to false
find media.cache_readahead_limit and change it to 9999
find media.cache_resume_threshold and change it to 9999
additionally if you'd prefer mp4 to webm
also in about:config, find:
media.encoder.webm.enabled
media.mediasource.webm.audio.enabled
media.mediasource.webm.enabled
media.webm.enabled
and toggle them all to false
note!
this will limit video to 1080p
and use https://addons.mozilla.org/en-US/firefox/addon/dont-accept-webp/ to kill WebP
Fuck Google
this sounds a lot like Y from another slavic language I know which means X in mine so this almost 100% also means X
this word is completely different but I can kinda tell the meaning from the morphology
what
okay right now I'm gonna have to dissect the entirety of this language's history to figure out how the FUCK did y'all get to the point of calling X that and not something more normal
I wanted to ask for X and accidentally called someone a whore
This reminds me of that time my (Czech) family went to Poland for my uncle's wedding (because my aunt - his wife - is Polish and they decided to get married in Poland).
The wedding afterparty was in full swing, everyone was drinking and partying.... and an elderly Polish lady approached my two aunts (who were also drinking a lot and partying hard) and commented something along the lines of "jesteście odporne". Which in Polish means "you're resilient" (as in they're handling the alcohol well) but in Czech it means "you're disgusting".
It took my aunts a bit to remember that it means something different in Polish. 😅
'pozor' in Czech means 'watch out!', but means 'shame' in Russian
droga in czech means drug, in polish it means road
záchod in czech means toilet, zachód in polish means 'west'
odchod 'to depart' in Czech means very specifically 'to depart by foot', while in Slovak it has a bit wider meaning, and buses and trains also combine with 'odchod'. To a Czech speaker this gives the funny implication that departing trains grow legs and walk out of the train station
Croatian 'mrdati' = to wiggle, means 'to have sex' in Czech
'Čerstvý, voňavý chléb' is Czech for 'fresh, nicely smelling bread'. In means the opposite in some slavic languages 'stale, stinky bread'. This kind of 'good meaning in one but bad meaning in other' duality happens surprisingly often actually, see also úžas and odporný above
otrok in Czech means 'slave' but in Slovenian and perhaps other South slavic languages it means 'child'. Very funny when combined with bazaar, turns a perfectly fine baby bazar into a slave bazaar
My friend has a fun sentence from Slovenian - "Hitri otroci kuhali jed na Dunaji". Literally none of the words here mean what a Czech thinks they mean. No not even Dunaj. Czechs read it as 'smart slaves were disemboweling poison by the Donau river', original Slovenian meaning is 'fast children were cooking food in Vienna'
Edit: wait one more. "srpanj" in Croatian means July, "srpen" in Czech means August. Listopad in Croatian means 'October', in Czech means 'November'. rujan in Croatian means 'September', říjen in Czech means 'October'. kwiecień in Polish means 'April', kviteń in Ukrainian also means 'April', květen in Czech means 'May'. Months are fucked up.
I Sexually Identify as an Attack Helicopter was one of the best works of sci-fi of our generation and one of the best works of transgender fiction ever written, and there are world renowned authors who still have successful careers after they publicly assassinated the nascent woman who wrote it. I don't think they should ever know peace.
I think this constantly and then I get angry for thinking it, because trans women should not have to be martyrs or saints to animate our politics and our art. that work should have been her debut, not her epitaph. I should be moved by her career, not her absence. I could spit.
Peglar Papers Real which means one of the only things left of him besides his body is a journal with lines reminding him who he was. Remembering who he was. And they thought it wasn’t him at first. But it was. It is!
Current twitter drama is Europeans confidently declaring that they don't need to drive or use overpriced public transport to get to the MetLife stadium for the World Cup; they will simply walk down the highway to get there. Girl it's New Jersey. They're gonna splatter you for fun.
If you manage to get on the turnpike before the cops stop you, a soccer mom is gonna do the Jersey slide in a RAV4 and turn your entire group into a wet speedbump
Image me gently taking your hand as I tell you the following:
This is ABSOLUTELY a perfectly fine footpath.
In fact, with how much space I’m seeing here, it is entirely plausible, that the European hordes will just create a temporary little Wanderweg right next to the highway. With that much space they might not even have to interfere with traffic.
But also have you seen the space between your highways? I‘d say the gaps each easily fit another whole stream of European walking hordes. Or maybe even two going opposite directions :D
tightly grips your hand with both of my sweaty hands.
the grass is not a permanent feature. there just happens to be a chunk of it there. the side of the road can vary from grass to swampwater ditches to steep embankments to absolutely nothing within a very short distance.
they will call in every highway enforcer in the state and mass incarcerate the lot of you before they allow you to Darwin Award yourselves across 6 lanes of traffic into an international incident or, (their real concern) impede the progress of the single most important north-south interstate corridor in america, ball kicky game be damned.
(I'm starting to believe that a fair number of you in the notes have We're Better Than Stupid Americans embedded so deeply into your cultural identity that you will Just Not Listen to anything we say to you about the material circumstances about the place we live in, rather than taking us at our word that there is a reason that most Americans travel the way we do and it's a good reason.)
discoholicmusic
pixelskylines
areudirty
2087
stormborns
theartofmadeline
izzy806
