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.
Happy happy birthday, dear friend! I hope your day is fun and relaxing, with cake, prosecco, and hockey boys books šššššš¾š„šš
Thank you! Youāve hit on all my plans for the day. š
It was vague nonsense, which was why he complained for the entire queue while Pansy dragged him towards āthe most gifted seer in Europe.ā
āYou need distracting,ā Pansy said.
What he needed, though neither of them said it, was Harry Potter not leaving for Australia in three days. Harry Potter, who smiled at him over Ministry reports. Harry Potter, whom Draco had been meaning to tell he loved for six months, two weeks, and one catastrophic almost-kiss.
When the seer peered into her crystal ball, Draco asked, dryly, āSo, whoās my forever partner?ā
āA man with spectacles.ā
Draco rolled his eyes. āRevolutionary.ā
āGreen eyes, dark hair, and a most interesting scar on his forehead.ā
That night, Draco wrote an owl before cowardice could return.
Harry came over instead of writing back.
ā
āAnd that,ā Draco said years later, āis how I married your father.ā
Their children shrieked with delight and tore back into the garden, apparently satisfied at the answer to their question.
Harry waited until they were gone, then said, āYou do realise she was probably a fake, and she read the rumours in the papers.ā
āObviously, darling.ā
Harryās smile softened. āThen why did you believe her?ā
Draco looked out at the garden, where their children were laughing and playing, all wild hair and muddy knees.
āI didnāt,ā Draco said, lacing their fingers together. āBut Pansy paid her twenty Galleons to say it, and I was too desperate not to listen.ā
Darcyās introduction in Pride and Prejudice is really āwhat if you had just had the worst month of your life because your ex-bestie tried to lover boy scam your baby sister out of her share of your dadās life insurance and your friend dragged you to a shitty party in a dive bar in the neighbourhood where heād just signed a short term lease, and you decided to let your bad mood show because you were never going to see any of the assholes in this stupid shitty bar EVER again. And your friend ended up making out with a girl heād just met there while you were stuck talking to her sister who was less cute and then her mother appeared and started trying to matchmake and started saying how if she was twenty years younger sheād clime you like a redwood and ooooh is that a black Amex, guess the next round is on you hahhahahahaha, while her other sister (how many fucking sisters does she have?!) flashed an obviously fake ID at the bar and ordered six vodka-diet red bulls and no one in her family except the less-cute sister even tried to stop her. And you went home and consoled yourself that you would never see any of these people again but then you met them over and over again because they live next door and your friend and the cute sister keep meeting up to make out but not actually date and then. You fall in love with the less-cute sister because it turns out sheās really witty and charismatic but she already knows and remembers and resents the fact that on a day when you were in a shitty mood you called her mid out loud in a dive bar.ā
okay but anecdotes about other people accidentally having other people's service dogs task for them because dog just went "oh you are doing thing i am meant to fix" has me laughing imagining shane just having a high anxiety day for some reason and someone's anxiety service dog (maybe still in training, so a lil confused still) continuing to try and approach and task, but shane is just CONVINCED that it's ilya somehow trying to get this dog to keep coming over, and he's already on edge anyway, so people noticing that this dog keeps trying to come closer just keeps winding him up more, which just makes the dog try to come back over, which just makes him tell ilya to knock it off again, which makes ilya say he's not doing anything, which makes-
and finally whoever the handler is just "sorry, she's still in training." and ilya is just "ah, a professional in training. what a smart girl." "yeah, she's training to be an anxiety service dog. :)"
and ilya just looks RIGHT at shane, who is suddenly
oh my GOD, they had been fighting about getting another dog because ilya wanted anya to have a sibling and shane wanted to call it at one dog because what if the next one ISN'T good and now they're going to have two dogs AND one is a problem-
so they're already miffed at each other, and shane is convinced ilya keeps somehow signaling this dog to come over because he's being an asshole about their fight about dogs, the exchange about her being an anxiety service dog happens, shane avoids eye contact COMPLETELY and manages to fully escape standing next to ilya
but then ends up retreating to an empty conference room that the dog and handler also happen to go to as a break place for the dog, and the handler is like no worries if not it's okay but do you mind letting her practice tasking? it's really okay if not but she seems really eager-
and what is shane going to do? say no i won't help out with training this dog? no.
but now he's laying down under the very reassuring pressure of this dog, totally physically relaxed, but just sighs and goes "...goddamnit" because he knows ilya is going to find out about this SOMEHOW and then they're going to end up with a second fucking dog
in their household, she is dogtor rosie because she has a degree even if she doesn't use it (ilya made it, it hangs on the fridge), and when shane tries to appeal to ilya to help him call rosie off when she's tasking, ilya shakes his head and refuses
dogtor rosie is SO good at this! she's a little angel (and sometimes menace), and i adore her.
i can also maybe picture anya picking up the basics of some of these moves/behaviors after the two dogs have spent some time together and she's seen rosie receive positive attention/praise for doing so. (ilya is for sure providing that, even if shane himself is more unsure at first)
all of a sudden, anya is also now fetching things With Purpose when she senses that someone is upset/anxious. she doesn't have rosie's discernment or training, so she's not grabbing water bottles or prescriptions (even if they are for the wrong person), but she is bringing whatever random item feels right in the moment. she has the spirit!!!
it's very hard for shane to stay in his anxiety spiral for long with two concerned dogs insistently handing him ilya's medication and.... the tv remote? and well, if rosie is going to press herself across shane's lap, anya is going to do it too!!! she won't be left out of the cuddle pile when a packmate is in distress! and if rosie is going to be licking and nudging shane's hands every time he goes to pull at his hair, redirecting him, anya is doing the sameāshe's not going to miss out on any opportunity to be pet, either!
ANYA DOING HER RESIDENCY UNDER SUPERVISING PAWSICIAN DOGTOR ROSIE
also the idea of anya frantically looking around for an item to give shane when rosie goes to grab the empty pill bottle from the counter because she WILL go looking for it if they don't leave it there for her is SO cute and funny to me. incomplete list of items anya has prescribed shane to make him feel better:
his sunglasses
ilya's keys
yuna's purse
a pair of tongs for the grill
her leash AND rosie's leash
a pillow from their bedroom
her stuffed hockey stick toy
the empty pill bottle when she got there first, prompting an annoyed rosie to try and take it from her because that's HER item!! give it!! meaning it actually did work because shane is now having to ref a sibling squabble over helping him with an empty bottle of medication that wasn't even his to start with
and @hardlyhinged adding your tags here because Best Wiggly Girl hopping the boards and getting carried off with a puck in her mouth is SO delightful to me
another update: Amal El -Mohtar wrote a small article on her blog (https://amalelmohtar.com/i-tried-to-title-this-post-for-twenty-minutes-and-failed/), one which contains the words ā[ā¦] and the upshot of it all is that corporate marketing people at Simon & Schuster now know the name Bigolas Dickolas.ā
I know it made it up to #6 at one point. Being a pre-existing fan of both Trigun and This is How You Loose the Time War made this especially enjoyable.
LMFAOO YEAH I JUST SAW IT HIT #3 been trying to keep this post updated but didnāt want to spam, and the speed at which these things have been happening is beautifully astounding so:
trigun, following bigolas dickolas, has also become a trending topic in business and finance
Interview with Bun (or Mr. Bigolas)
Crossover art (https://twitter.com/xxrat__punkxx/status/1656446745088278529?s=20)
thread of semi-current placements on the book
memes
and the situation very much escaping online confinement xD
A significant number of hours later (and too much to summarize):
Time War summary: reached #3 on Amazon overall, #1 (and 2, And 3) in science fiction romance, #3 in lit and fiction, social media presence overseas, and-Ā
printing with a bigolas dickolas sleeve and so much more
Bridging the two: so many articles
slate article
techcrunch article
insider article
kotaku article
Gizmodo article
tor.com
polygon
and etc
And then on the bigolas dickolas trigun side-
And then after tweeting @ dark horse for a potential trigun maximum (trimax) reprint due to the eng verison being out of print for several years by this point:
so everything can now basically be summed by by this tweet!
(I donāt know if iāll keep updating this thread considering how many branches there are and idk how much more this can escalate, but Iām keeping an eye on the situation so weāll see!!)
For those asking what time war is about, per Mr. Bigolasās post: DO NOT look up anything about it. Just read it.
For those asking about trigun, it is an amazing manga, anime adaptation, and new reboot anime.
its been over a year since all this first occurred, but felt it was necessary to tack on that bigolas dickolas declined the Hugo that the infamous tweet was nominated for š
jbbartram-illu
jacobvanloon
annaxmalina
theartofmadeline
turnnoffyourmind
discoholicmusic
