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.
One of the other reasons I've really come around to the midi-chlorians as a concept (the Force being interactions between magic space bugs that have flooded this particular galaxy and every living thing in it) is that my family has long had many discussions about the midi-chlorians potentially also "explaining" droids and the computer situation.
Because SW computers are apparently incredibly simple and/or one-purpose (as a result of the original movie having been made in the 70s), but droids are incredibly complex, capable, independent, remarkably emotional, irrational, free-willed, etc., compared to other technology in this universe. SW writers can't seem to decide whether or not droids are fairly basic machines or fully sapient beings.
So, one of our solution headcanons has generally been that: midi-chlorians REALLY like feeding on or making their home in various technological materials required for computers. Any sufficiently advanced technology BECOMES magical, whether you want it to or not, because the magic space bugs really like infesting these machines. And since the magic space bugs hold the memories of the cycle of life of countless living beings in this galaxy, etc., you could snappily phrase the inevitable sapience of complex computers in this particular galaxy as: "Droids are accidental composite Force ghosts, actually."
So, in SW (in this headcanon), if you DON'T want your advanced technology to randomly develop a personality and capabilities beyond its programming, you have to design your complex computers BIG with all of these bulky, heavy anti-midi-chlorian safely measures. Maybe tech factories are best built on 1) planets with no prior history of life (and thus midi-chlorians) or 2) planets the midi-chlorians hate for some reason, because that produces the most reliable products, and you can try to seal up a computer or a droid's systems as best you can before sending them out into the magic-bug-flooded galaxy.
So, you can have reliable, non-sapient complex computers in big space stations or communications stations or big ships, but that's not very portable! If you want advanced portable tech, you need a droid, and the midi-chlorians are almost inevitably going to get in there (especially once any repairs are done by cracking them open) and get comfortable (it's SO hard to get them out fully again), and so your space navigation machine or translation device is inevitably going to develop 1) sapience, 2) a personality, and 3) some kind of mental illness because droids are inherently pretty diverse, unstable lifeforms put to difficult and unpleasant purposes. One way of dealing with the inevitable infestation is just to design your droids accepting that they will develop some form of sentience &/ sapience, and putting in some safety rails in the hardware and software accordingly.
Because of this, being an IT person in the SW galaxy sucks even more than usual, because someone will call you up complaining that their household terminal is acting up, and you have to try to figure out if and how to explain that the computer has started running itself partially on an infestation of microscopic magical space bugs and is inventing new processes and programming accordingly. So, the solutions are basically: 1) wipe most of the system, replace a few key parts, and accept the loss (knowing that this will happen again in a few years), 2) buy a brand new machine and be REALLY careful about any file transfers if any, OR 3) give the computer some cheap wheels and a way to communicate to see if that makes the baby ghost now possessing the machine calm down enough to be persuaded to keep doing the damn job.
Because they can get very hot and then burst, and if any of the very hot filling material gets on your skin, it will burn you. The filling material is also usually sticky, and you do NOT want burning-hot goop to get stuck to your skin (source: USCPSC)
You also shouldn’t microwave them, for the same reason.
If you own these, keep them away from heat sources (direct sunlight, the stove, your radiator, etc).
I love it when a classicist has clearly gotten so lost in the BCE sauce that their perception of time has become deeply warped. Case in point: the article I’m currently reading just described the Byzantine era as “a period relatively modern.”
Usually when the electrician goes 'hey.... there's something you should see' it's because someone set up shoddy wiring. This time it was because he pulled two snakes out of my wall
A doll youtuber I watch has made a video about the history of Polly Pocket, and she's describing the plots of the dvd specials, one of which features an elderly woman named Ms. Throckmorton, and my reaction was
one time i was making potato soup and i needed broth for it but i was out of broth. but in my dog food cabinet there was a carton of chicken broth that i had bought for my dog. now the box did say HUMAN GRADE and the ingredients were just bog standard chicken broth. it did say something about joint health for dogs on there. but the ingredients were normal. i even compared the ingredients list to some normal person broth and they were the same. so i did put dog broth in that soup and it did taste completely normal. but i did use the dog broth.
psychic beam blast. none of you understand the late victorian eugenics/determinism/orientalism/scientific history/racial-sexual-class great chain of being unity of affect enough to be talking about what dracula does and does not represent
dracula serving as a synecdoche for european aristocracy is in no way at odds with his widely recognized status as a gross antisemitic stereotype, because in the late victorian (especially english) literary mind medieval aristocracy = regressive = disabled/malformed = irrational = feminine = catholic = eastern = jewish = romani = queer = underevolved, which in turn smacks of irishness and spirituality. these are all linked together—explicitly in the scientific and literary worlds of the 1890s and more covertly today—in a web of emotional associations that smooth over all contradictions, and all adds up to: we have moved into the rational, masculine, blue-eyed light called modernity, but something is clinging to our ankles and wants to drag us back into the dark. are you one of the things on my ankle? prove to me that you are not. step into the light and assume your proper place under me, or i will defend myself by slicing you away
A funny, bordering on charming example of people knowing about Octavia Butler but not actually reading her is that I saw a picture of a book display labeled, “they saw it coming” with several of her books on it, including the one where earth is taken over by tentacled aliens who make humans be in polycules with them. And friends, either we were, in creating this display, working off of a vague sense of what black ladies write about or there’s something I’ve missed about modern family structure since I have noticed zero tentacled throuples in my day to day life.
I have just enough knowledge about archaeologists working for the American federal government to write the most boring stargate: SG1 fanfiction ever put to page. Everyone be thankful I have not inflicted it upon you.
It would also be a good opportunity to set the fanon record straight a bit. Nobody in the archaeology world would be shocked about an archaeologist working for the Air Force. Working for the federal government is one of the most normal career trajectories for archaeologists in the United States. If you search “archeologist” (note, the federal spelling omits the second ‘a’ for some bureaucratic reason) on USAJobs right now, 5 open positions come up. Two of them for the army.
Granted, Daniel’s a bit of an odd case since he got run out of academic archaeology first instead of coming up via the CRM track, but still.
“Why would the Air Force hire an archaeologist?” Is the wrong question. They (and every other branch of the military) have dozens of others for completely normal non-aliens reasons. (And no, im not talking about contractors. People performing Inherently Governmental Activities are gonna be civilian GS employees.) Some of them attend SAA events every year perfectly normally. Better question: why has nobody from the Colorado State Historic Preservation Office, the Office of the State Archaeologist, or the Ute tribe met this guy?
#basically there are a lot of people out there who think they know what Daniel’s job is #and also think he’s terrible at it #‘What do you mean Cheyenne mountain has had an archaeologist for TEN YEARS and he’s never emailed us once?’ #<- statements overheard at SHPO
Okay but that is like, a missed opportunity for a filler episode: Daniel’s job in-between missions is to figure out how to fake just enough that he can publish papers and attend conferences that the outside world doesn’t get suspicious of the Stargate Program.
Like, he meets with other archeologists on earth and comes oh so close to blowing his cover time and time again when he almost corrects them on some minutiae they’re publishing about because he’s met people from that culture who are alive and well…but he can’t really let his argument be that, so he just flails.
He’s still a bit of an outcast, because everyone sees him as having wild theories he can’t back up when in reality his findings on ancient cultures would alter sociological understandings like crazy.
gummygunk
jimmyeatworldofficial
blakekathryn
pixelskylines
