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.
Hey, man, c'mere. Listen. Get in real close, this is important.
You're gonna make stuff again. You're gonna make stuff you're proud of. You're gonna make stuff you're excited to share. You're going to feel that overwhelming drive to create, not just the frantic I want to want to you're stuck in now. You're going to have awesome ideas, and you're going to make them into reality. You're going to create again. You're still an artist. You're still a writer. You're still home to the same passion you had before. You'll find it again. It's not gone. It's just resting. Let it rest. You're going to make stuff again. I promise.
Workers should have the legal right to tell their boss to fuck off and refuse to come into work for an indefinite period of time to recover from burnout and then be able to come back to work 5 weeks later as if nothing happened.
Tbh germ theory DOES sound crazy. Like if you told a regency-era nobleman that tiny creatures lived on the surface of everything and THAT’S what causes consumption, they’d be like “ah, I see you are a lunatic. Would you reside in my hermitage? Rantings and ravings do so amuse my guests”
Yeah, actually, it would probably be pretty easy to explain germ theory to a Medieval person as tiny evil spirits that live on everything, but they can be purified by soap and water, or by alcohol, because that is why God has granted us those things. And because they can float in the air, if you cough or sneeze after they have infested you, that can cause them to infest others. And when you are sick, the angels God has deputized to defend the bodies of His beloved children are at war with the evil spirits, and, sadly, sometimes they lose, but the best way to help your angels win their battle is to rest, drink plenty (this would probably be small beer in this time period, not water, because the water was also infested), stay clean, and for the sake of God do not allow anyone to let your blood, for the angels need that blood in their war against the evil spirits. Bloodletting is good for some types of illnesses but not the kinds caused by the tiny evil spirits.
boiling as a sterilization measure is also easy to explain. water returns to the air when heated and it rises as steam back up to the floodgates of heaven; we know God created the world in seven days, He's not up there making more water every time it rains. it circulates. the returning of water to heaven also purifies the water of unclean and malign influences. you know wormy water from a muddy puddle will kill your kid. you know you wouldn't wade into a bog and have a slurp. water that remains in the low places of earth absorbs all that is unclean from our waste and it may also sponge up new diseases from hell, we're not totally sure about that one, but it seems likely. God set up the heavenly water cycle so that the earth's waters wouldn't totally fill up with gunk.
what does this have to do with boiling your surgical tools? well look, the boiling water releases bubbles of steam which carries the malign influences up to heaven. you boil a knife, you send all the miasmic particles off with the steam to heaven. if you rinse the knife off in a bucket the water isn't hot enough, the particles go into the water and then right back on to the knife. you gotta boil it to get the particles all the way away.
how can a tool or rag or a bed have miasmic particles on it when you can't smell them? humans have a lousy sense of smell. look at your dog on the hunt. are there no rabbits in the woods just because you can't smell them? we know that miasma is carried on the air, and is what makes stench so dangerous, and we know that humans can't smell worth a damn compared to dogs cats horses etc. a dog can smell if a rat died in a corner of the room last week. you can't. do you think licking the spot where the rat died is going to go well for you? luckily, what humans lack in snout we make up for in brains. we have extra brains where our sniffers should have been. God set that up for a reason.
and why does a rinse with wine spirits work? man, look how fast alcohol evaporates. my guess is that because wine contains a lot more vice than water, it evaporates a whole lot faster, in sort of an equal and opposite way that a rock falls faster than a feather. if you want the miasmic particles to get off there FAST, you dunk it in something that's going back to heaven at a gallop.
what's up with honey? it just preserves things against corruption. doesn't clean them off. honey doesn't evaporate at all. probably because bees don't sin. it's not good for ridding a tool of particles-- it's sticky-- but fine for preserving anything you don't want to go to heaven OR hell. this is why you wash the wound with wine spirits or purified water FIRST, to sluice the miasma out, then slap the honey on AFTER. and boil the damn bandage, too. you wouldn't put a rotten door in a sound doorframe and expect it to keep out bandits, would you? cmon.
something i find really interesting about (some) people who don't write is the way they'll elevate plot over execution? it's the way you'll sometimes see people who think they have a best-selling book idea, and they want someone to write it for them while thinking they should get the lion's share of the imaginary profits, because it's their idea. and don't get me wrong, ideas are important! plots are important!
but there are so many stories i've read over the years that could be summarized into a few sentences of bland-sounding plot, and it's the execution that matters. it's the writing. it's the writing!
a talented writer can turn a bland plot into a story you want to devour. and someone who only cares about ideas can take the most interesting idea ever conceived and make it unreadable/unwatchable.
like it does feel like there are a Lot of people in this world who don't view writing as a skill that deserves respect, and who think that really it's all about coming up with shiny ideas. and it does feel like that's uhhh infected a LOT of different things in the modern world. tbh.
i hope you buy things that bring you joy. i hope you wear clothes that make you feel good. i hope you decorate your home however you like. i hope you don’t wake up dreading going to work. i hope you meet people that make your heart sing. i hope you discover hobbies that fulfill you. i hope you go on fantastic adventures that give you unforgettable memories. i hope you do things just because you want to, and i hope you don’t feel like you have to justify your decisions to anybody. i hope you find happiness. i hope you cultivate a life you love
My hope for whoever is reading this is that your life starts making sense and coming together. I hope the good days are right around the corner for you.
being at work while your personal life is falling apart has to be among the top 3 worst human experiences. You’re at your absolute lowest and someone wants to circle back on an email…… unreal
2087
turnnoffyourmind
izzy806
yousharknotpass
jacobvanloon
