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.
hereās a picture of a baby cedar waxwing begging for food from a robin. neither of these species are nest parasites, so itās not possible the cedar waxwing was āadoptedā. this is essentially the bird version of tapping a random person on the shoulder at the grocery store and going āMOMā
Robin is bathing which makeās it even funnier. More like the equivalent of a strange kid barging in while youāre taking a shower and demanding you make them mac n cheese right now
my toxic xennial trait is that i believe something should either be software (in which case after i download it i shouldn't need to be connected to use it) or a web page (which shouldn't require me to download anything to use it, however badly, in a browser). fuck your mandatory single function constant connection apps
You know as an American born in the late 90s Iāve watched my country blow up foreigners on tv supposedly in the interest of national security since I was a toddler but quite frankly we live on a natural fortress of a continent with friendly nations to our north and south and have been attacked on our own soil literally two times in the past hundred years and we have military bases on every inhabited continent on earth. Itās never made sense to do these things for ānational securityā. It never will make sense unless you value the shareholders of weapons companies over the lives of children. And apparently a concerning amount of people do.
they really don't make Bigass Cabinet For Computer anymore. we were at the used furniture warehouse the other day & they had so many. that's the way things should be. lock away that wretched machine when you're done looking at it
i think this particular gag has gone out of style but it really is always funny when movies make their soundtracks diegetic in situations where it makes absolutely no sense
Can I ask why the Pinkerton agents get such a bad rap online? All I know from them comes from the Sherlock Holmes novel The Valley of Fear, where they help take down a gang of extorsionist coal miners.
So the Valley of Fear concerns a real-life incident that helped to shape the Pinkertons' reputation: their destruction of the Molly Maguires in the coalfields of Pennsylvania. An early trade union of miners of mostly Irish extrction, the leadership of the Molly Maguires were accused of murder, arson, kidnapping, and a whole host of other crimes by undercover Pinkerton agents who had infiltrated their ranks on behalf of the Philadelphia and Reading Railroad, who the Molly Maguires were fighting over wages, hours, and working conditions. Dozens were sent to jail for long stretches of time, and ten men were ultimately hanged as a result of Pinkerton testimony.
The fact that the evidence against the Maguires largely stemmed from undercover informants paid by management to disrupt and destroy their organization and other suspect sources of evidence like jaihouse snitches, and that the prosecutions against the Maguires were personally carried out by the CEO of the Philadelphia and Reading Railroad in the midst of labor conflict over who ran the coalfields, has led a lot of labor historians to conclude that the Molly Maguires were stitched up in the courts and that the Pinkertons committed wholesale perjury, acted as agents provocateur to provoke the crimes they then testified to, and caused the judicial murder of ten men.
That was just the start of their long and inglorious history of being the favorite goons of anti-union robber barons: the gun thugs who were outfought at Homestead were Pinkertons, as were the private armies who fought for capital during the Great Railroad Strike of 1877 who had to be once again bailed out by the National Guard. Likewise, their foray into bearing false witness continued with the manufactured confession that named UMW leader Big Bill Haywood as the man who arranged for the assassination of Governor Frank Steunenberg of Idaho.
So yeah, the Pinkertons have earned every last drop of their evil reputation. There is blood on their hands.
oh my god i'm so tired psychotic does not mean violent it does not mean angry or erratic. it refers to a person suffering from psychosis, a loss of touch with reality that includes hallucinations and/or delusions. psychotic people are not inherently violent and y'all need to understand how much stigma you create when you again and again incorrectly use the word psychotic without even thinking about it
my rule of thumb for this, for any behavior that you look at and go, "god, how can you not (X)?" is to ask myself: well. how bad would it have to be, for me? how bad would it be before i would stop doing that thing? how bad would i have to feel?
what would have to go wrong in your life, and how wrong would it have to go, for you to stop bathing? to stop eating? to let the garbage pile around you until you can no longer see the floor?
how bad would you have to feel, and for how long, before you would stand on the street screaming at anyone & anything? beating your fists on your head and crying? how bad would it need to be?
do you think there's some fundamental difference between you and people who suffer until their lives are unrecognizable to you? can you face the knowledge that if things went badly enough, you would be just like them? just like them? just like
Thought crimes aren't real, telepaths don't exist, and even if they were and do, having your head constantly play "Fly me to the moon/Let me kick its fucking ass/Let me show you what I learned/In my moon jiu-jitsu class" on repeat while out and about in public isn't illegal
ellievsbear
izzy806
discoholicmusic
yousharknotpass
areudirty
