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.
By Philip Jackson. I did not find all of them in his gallery, but 1. Moonstruck, 3. The Sentinels, 4. The Grandees, 7. His Eminence, 8. The Magistrate, 9. Saraband
Seattle Central College is trying to close the only public wood tech college in the state of Washington, which is one of if not the only avenue into woodworking trades for people who usually aren't hired to be trained on-site: queer, trans, neurodivergent and disabled students, and women. Which is exactly what most of the students in the program are.
The reason according to the college is that they want to sell the building to fix the school's budget; they also claim the wood tech college has a severe budget deficit, but they've failed to follow through on any promises to try to fix that (including hiring a permanent dean, so the school has someone to advocate for them... which they haven't had in at least a decade).
The reason according to some students & staff is that Seattle College's VP has a grudge against the lead carpentry instructor & has been trying to get her program shut down for years to get rid of her.
Students are trying to make some noise about it, in hopes that the college backs down and they can work towards the long-term fixes Seattle College keeps failing to follow through on.
The future is unclear for one of the Puget Sound region’s only comprehensive training programs for carpentry and boatbuilding.
If you can sign this little petition & share this around, it'd help the school out a ton!
It shows the school they're going to get pushback, which has changed their mind about this in the past.
It would be really cool if we could help get this to 1k!
There are less than 200 students at the wood tech college, so makes a really big impact when anyone else signs and shows that people outside of just students really care about this.
There were like 230 signatures when I first posted this- I'm really grateful to everyone who's already helped out. It means a lot!
”I have this artistic idea but not the skills to achieve it to the standard I want.”
congrats! Now you have a motif! A recurring theme! A focus for your art! Something to haunt you!
Seventeen still lives of dandelions? Three hundred poems about grief? A sketchbook dedicated to your grandmother’s house? Two books trying to unravel the complexities of familial relationships?
Don’t let the fear of it not being perfect on the first try stop you from being Weird About It!
Please view Hokusai's gradual working towards The Great Wave Off Kanagawa, over a period of 39 years.
An early exploration of the themes Hokusai would keep coming back to is Spring in Enoshima, done in 1793 when he was 33. The wave is small and there are no boats, but Mt Fuji is clear in the background, and Enoshima is in Kanagawa, so we are clearly beginning to work towards something here.
A second pass, eleven years later in 1803 when he was 44. The title of this one begins to get more familiar: The View of Honmoku Off Kanazawa. It has a towering wave over a smaller boat, but Mt Fuji is not present, and the boat is considerably larger and has a sail. But the feeling of danger in the wave and the smallness of the boat are here, and of course the general composition is definitely recognizable.
This is A View Of Express Delivery Boats, done in 1805, merely two years later at age 46. Here we find the wave and the boats almost exactly as we'll find them in The Great Wave Off Kanagawa, though Mt Fuji isn't present, and the location is uncertain. And it's a good picture! The wave is threatening, the boats are small -- but the feeling of "ocean" isn't really there yet, is it? It's unlikely this picture would have become a classic for the ages. But that's okay, there's still time.
And here we have it, a full 26 years later, done by Hokusai in 1831 at the age of 72. The Great Wave Off Kanagawa, one of the most recognizable pieces of art in the world. The boats are there, the mountain is there, the wave is there, and the FEELING is there. He did it! He reached the apex of his ongoing motif and theme!
Or did he? Because the whole point of a motif is not that you're striving to get to the perfect version of it, the one idealized image you carried in your head all along, and when it is done, you are also done. Hokusai is on record at the age of 73 saying he'd only just begun to feel like he was learning how to draw things properly, and that "if I keep up my efforts, I will have even a better understanding when I was 80 and by 90 will have penetrated to the heart of things. At 100, I may reach a level of divine understanding, and if I live decades beyond that, everything I paint — dot and line — will be alive." He had drawn The Great Wave, but he didn't believe he was finished -- he thought that he was still just beginning to get started.
And he wasn't finished with his ocean motif, either. Please check out his Mt Fuji At Sea, done in 1834 at the age of 75.
It's all there; Mt Fuji, the ocean, the wave. The boats are gone, but replaced with birds, flying with the wave instead of fighting against it. It's not as famous as The Great Wave Off Kanagawa, but that's not what motifs are for -- each successive work does not have to surpass the previous in terms of success, especially in terms of external success. They're there for you to keep playing with, keep remixing and re-experiencing, for as long as you think you have something to say.
I also want everybody to know that Google and most of the internet think that all of those paintings bar the last one are called "The Great Wave Off Kanagawa", so I had to do a sort of middling deep dive just to find their actual names. And then I was like "I don't think those translations are very accurate", so I went on a second quest to retranslate them, which was particularly difficult with painting three (A View Of Express Delivery Boats) because for some reason he titled that one entirely in hiragana, and it's all archaic words that were very hard to chase down without their corresponding kanji. Google suggested "the push-off is a transportation route", which wasn't particularly helpful.
All of which is to say that I probably spent a bit too much time on all of that, but it was fun; and at least I know what those paintings are called now.
I’ve been slowly working on illustrating my own Tarot deck about black cats.
I have a long way to go, but I’m nearly done with the major arcana.
Here are some of my favorites:
I set myself a design challenge (as I often do). I wanted to limit the color palette and the brush styles to create a very consistent look.
And since each card focuses on black cats, it forces me to think deeply about symbolism and design to make sure the meaning of each card is clear and each design is unique.
But there’s still a lot of variety coming through. It’s not all class, or all silliness - just like cats
I have a forearm tattoo of the cats in this one ☝️
I’m so happy with how both of these two managed to feel dark, but adorable at the same time.
And then there’s this one absolute diva:
Most of these used my own feline friends for reference.
There are more but I don’t know yet how many I want to share online. If I can actually complete the entire deck I would love to have it made. But I have a ways to go, so for now I continue to chip away, one card at a time.
i have this painting i got at a yard sale for free (if i took a broken dresser with it, which i did) and i have been obsessed with it:
so today i was trying to figure out where it was from and the more i searched the weirder it got because near as i can tell it's from the fex valley in the swiss alps which is cool but what kinda chokehold did this specific bridge/cabin have on people
these are two different paintings... and then i found this postcard on ebay
all these ppl painting the same exact image from the exact same random postcard... anyway if anyone knows this spot i would be fascinated to know more but i will live in mystery and if you're wondering "did you buy the postcard" yeah of course i bought the postcard
okay thank you @thepinkpeg who did find this and it turns out it's actually a 17th century cabin now converted into an art exhibit accessed only by foot (or horse). from the outside on the hill above it looks like this:
and inside it contains this:
thank you all so much. i have my answer and it is more fascinating than i could have thought.
hello fellow artists. google has fallen. pinterest/duckduckgo AI filters don't work. do not despair; here is a list i made of places to find reference images without having to sift through piles of worthless garbage. (for future editing convenience i am just linking my blog post on dreamwidth.)
✨ good places to find art reference that are not full of AI trash 🌈
imagine if the oceans were replaced by forests and if you went into the forest the trees would get taller the deeper you went and there’d be thousands of undiscovered species and you could effectively walk across the ocean but the deeper you went, the darker it would be and the animals would get progressively scarier and more dangerous and instead of whales there’d be giant deer and just wow
#one day the idea that spiralled from this concept will be more than a nebulous blob in my brain #promise #it's only been years #fragments of the first #inspo
I made some weird little watercolor and gouache paintings for a sci-fi-themed exhibition. I was miles out of my comfort zone and some of my paintings were Quite Bad but I had fun
tangentwithtanner
prguitarman
germansingergirl
janainaart
romaincrisis
ellievsbear
stormborns
