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.
the best thing about the project hail mary book that was completely missing from the movie is Ryland Grace constantly complaining about how being an American scientist he's constantly dealing with different incompatable units of measurement and it's good to know an American - even a fictional one - understands what it's like to be any Canadian ever
Loving how this low key implies that Spike would spend years trying to best Miss Piggy in a fight, would then secretly fall madly in love with Miss Piggy, have an frenemies with benefits situation with Miss Piggy, and then ultimately sacrifice himself to save Miss Piggy's life because that is the only way Spike knows how to deal with a blonde bombshell nemesis. He has no other option.
I'm starting to gain insight into why people turn into conspiracy theorists. Some topics are so totally neglected that it looks like they were intentionally and maliciously erased, instead of falling victim to arbitrary lack of interest.
I think it's a vicious cycle; when people don't know something exists, they're not curious about it. Also, people use conceptual categories to think about things, and when a topic falls between or outside of conceptual categories, it can end up totally omitted from our awareness even though it very much exists and is important.
This post is about native bamboo in the United States and the fact that miles-wide tracts of the American Southeast used to be covered in bamboo forests
@icannotgetoverbirds It already is a maddening, bizarre research hole that I have been down for the past few weeks.
Basically, I learned that we have native bamboo, that it once formed an ecosystem called the canebrake that is now critically endangered. The Southeastern USA used to be full of these bamboo thickets that could stretch for miles, but now the bamboo only exists in isolated patches
And THEN.
I realized that there is a little fragment of a canebrake literally in my neighborhood.
HI I AM NOW OBSESSED WITH THIS.
I did not realize the significance until I showed a picture to the ecologist where i work and his reaction was "Whoa! That is BIG."
Apparently extant stands of river cane are mostly just...little sparse thickety patches in forest undergrowth. This patch is about a quarter acre monotypic stand, and about ten years old.
I dive down the Research Hole(tm). Everything new I learn is wilder. Giant river cane mainly reproduces asexually. It only flowers every few decades and the entire clonal colony often dies after it flowers. Seeds often aren't viable.
It's barely been studied enough to determine its ecological significance, but there are five butterfly species and SEVEN moth species dependent on river cane. Many of these should probably be listed as endangered but there's not enough research
There's a species of CRITICALLY ENDANGERED PITCHER PLANT found in canebrakes that only still remains in TWO SPECIFIC COUNTIES IN ALABAMA
Some gardening websites list its height as "over 6 feet" "Over 10 feet" There are living stands that are 30+ feet tall, historical records of it being over 40 feet tall or taller. COLONIAL WRITINGS TALK ABOUT CANES "AS THICK AS A MAN'S THIGH."
The interval between flowering is anyone's guess, and WHY it happens when it does is also anyone's guess. Some say 40-50 years, but there are records of it blooming in as little time as 3-15 years.
It is a miracle plant for filtering pollution. It absorbs 99% of groundwater nitrate contaminants. NINETY NINE PERCENT. It is also so ridiculously useful that it was a staple of Native American material culture everywhere it grew. Baskets! Fishing poles! Beds! Flutes! Mats! Blowguns! Arrows! You name it! You can even eat the young shoots and the seeds.
I took these pictures myself. This stuff in the bottom photo is ten feet tall if it's an inch.
Arundinaria itself is not currently listed as endangered, but I'm growing more and more convinced that it should be. The reports of seeds being usually unviable could suggest very low genetic diversity. You see, it grows in clonal colonies; every cane you see in that photo is probably a clone. The Southern Illinois University research project on it identified 140 individual sites in the surrounding region where it grows.
The question is, are those sites clonal colonies? If so, that's 140 individual PLANTS.
Also, the consistent low estimates of the size Arundinaria gigantea attains (6 feet?? really??) suggests that colonies either aren't living long enough to reach mature size or aren't healthy enough to grow as big as they are supposed to. I doubt we have any clue whatsoever about how its flowers are pollinated. We need to do some research IMMEDIATELY about how much genetic diversity remains in existing populations.
Many years ago I did some (non-academic) research on native canes in the USA because I thought I remembered seeing a bamboo-like something in the wild that I'd been told was native, and I thought it might make a nice landscaping accent. But the sources I found said something like "unlike Asian bamboos, the American equivilant barely reaches the height of a man", and I went "nah, that is exactly the wrong height for anything." But if it gets 10 feet and up, I think there are a lot of people who would be VERY happy to use it as a sight barrier in public and private landscaping, and if it means putting in a bit of a wetland/rain garden, all the better. The lack of a good native equivelant to bamboo is something I have heard numerous people bemoan. Obviously it's very important to protect wild sites and expand those, but if it'd be helpful, I bet it wouldn't be hard to convince landscapers to start new patches too.
For instance, a lot of housing developments, malls, etc. seem to set aside a percentage of their land for semi-wild artificial wetlands (drainage maybe?) planted with natives, and then block the messy view with walls of arbovitae or clump bamboo from asia - perhaps it would be a better option there?
Good Lord. Arundinaria isn't just a better option, it's perfect.
I was in the canebrake near my house again this morning, and river cane is extraordinarily good at completely blocking the view of anything beyond it. It is bushier and leafier than Asian bamboos, and birds like to build nests in it. It would make a fantastic privacy barrier.
The cane near my house is around 10-12 feet tall. This species can reach 30 feet or more, but I think it needs ideal conditions or to be part of a large colony with a robust system of rhizomes or something.
It grows slowly compared to Asian bamboos, and seems to need some shade to establish, so it would take time to become a good barrier, but no worse than those stupid arborvitae.
plants like this were often intentionally cultivated in planter boxes as a form of water filtration and civil engineering by a bunch of indigenous nations.
There's a reason why Native Americans cultivated canebrakes.
Well, several reasons. As y'all may know, bamboo is stronger than any wood, and therefore it makes a fantastic building material.
The Cherokee used, and still use, river cane to make fishing poles, fish traps, arrows, frames for structures, musical instruments, mats, pipes, and absolutely gorgeous double-woven baskets that can even hold water.
This stuff is, no joke, a viable alternative to plastic for a lot of things. The seeds and shoots are also edible.
Uh I know this is out of left field but I work in plant cloning - it's a lot easier than you'd think to do for plants and it's honestly a really important conservation tool, and good for making a TON of seedlings in a short amount of time. I can look into this genus for like, cloning viability?
Hi y'all, reblogging the Canebrake Post again. It's been over a year since I fell in love with the coolest plant ever. I'm trying to bring it back but I am very small so if any of y'all have a Canebrake nearby you might wanna talk to the owners and contact some local parks and nature preserves yeah?
A lot of people are asking how to distinguish Rivercane from invasive bamboo species. This link should help you!
Here's some distinguishing traits I've observed myself:
River cane has a really full, bushy, leafy look that makes it really hard to recognize as bamboo from a distance, because the stems are harder to see. The shape of the individual cane with its branches and leaves is narrow, because the branches spread out very little, but the foliage is DENSE. It's like a plume.
River cane is stronger, denser and heavier than invasive bamboos I've seen.
River cane stems are always green all the way around, no yellow (unless the plant's been dead for a good long time)
River cane stems feel smooth like plastic to the touch. The common invasive bamboo I've seen here, when you run your hand upwards along it, the stem feels awful like sandpaper.
The biggest way to distinguish them: River cane grows 6-4 feet tall when it's in little patches, and up to 10-12 feet when it's in a large size patch (like, the size of a backyard) It is known to reach up to 15 feet tall nowadays and historical records claim heights of 30 feet or more in fertile river valleys. I really want to stress that it's RARE for it to get big. A canebrake will almost always be many times wider than it is tall (sometimes they grow in very long strips along fence rows)
The best time to look for it is in winter before things leaf out, because it's evergreen and grows in dense masses, making it easy to spot.
Some more cool stuff i've found out—River cane was a common food of bison! Earliest European settlers reported canebrakes so big that "100 bison could graze on a single canebrake." Apparently it used to make extremely high quality forage for livestock, before it was mostly destroyed.
European settlers apparently set their pigs loose in the canebrakes purposefully to destroy them, because the pigs would root up the nutritious rhizomes and kill the plant. Thinking of the relationship between Bison and Canebrakes, and the relationship between Eastern Native Americans and Canebrakes, and the relationship between Plains Native Americans and Bison...it seems like a pattern, huh?
In the case of both bison and canebrakes, they were a fundamental part of their ecosystem, and fundamental part of the indigenous cultures that used them for every material, their musical instruments, their homes, their most advanced arts, and even food (Rivercane shoots are edible just like other bamboo, and supposedly the seeds are edible too!) but European settlers purposefully destroyed the species almost completely. I can't help but wonder if there was a similar motivation.
Books that talk about Rivercane:
Weaving New Worlds: Southeastern Cherokee Women and Their Basketry by Sarah H. Hill talks about rivercane a LOT and gives tons of details of its uses and history.
Saving the Wild South: The Fight for Native Plants on the Brink of Extinction by Georgann Eubanks has a whole chapter about Rivercane.
Venerable Trees: History, Biology and Conservation in the Bluegrass is a book about Kentucky, but it talks about rivercane's importance including its relationship with bison. It's only a couple pages out of the whole book but it's still great information.
By the way, though, if you read any very early European account of Kentucky, the word "cane" is everywhere. It's just such a nondescript word it's hard to realize its significance.
On a more personal note...god, I love this plant. Here's another photo I took. When you're in the canebrake, it feels so cut off from the rest of the world; it's shaded, quiet, cool, and someone 10 yards away couldn't even see you.
i actually talked to my neighbor that I learned owns the canebrake. She had no idea what it was but she was excited to learn about it! It was a lovely conversation.
Apparently, she knew I had been down there a bunch of times and thought nothing of it. She said "Yeah I told my husband, If you see her down there, just leave her alone she's doing her thing." In the most sincere way possible, God bless this woman
She said I could transplant all I wanted, too. This was great! ...but I quickly learned how RIDICULOUSLY HARD it is to transplant from a canebrake of this size. The rhizomes are so big and tough, a shovel can hardly get through them, and unless you're at the edge of the canebrake, there's a thick mat of them going every which way. I was driving my whole weight down on this shovel and it kept just denting the rhizome and glancing off.
I did get some transplants but each one took like half an hour because I was fighting for my life!
Also, with a canebrake this size, it doesn't grow little canes that will later become bigger—it shoots up tall canes in a single season. The youngest canes, more accessible and toward the edge of the canebrake, were significantly taller than I was. I cut the top off of one transplant for ease of handling—I had a pair of hand pruners with me that were usually perfectly useful for small limbs, but I could barely get these things through the cane, it's just so strong and dense.
Someone research the material properties of this stuff ASAP. It's insanely strong.
Here is some YouTube videos that talk about river cane!
Roger Cain of Keetoowah/Western Band Cherokee shows and talks about Rivercane. This video has a BIG canebrake, the mature canes look as if they could be 15ft tall, but he says it's only a fragment of what they used to be!
Stan the River Man visits a Canebrake in Northern Kentucky. This channel only has 22 subscribers, I feel like I've discovered a rare and priceless treasure
River Cane Renaissance, Episode 1. This guy has devoted a large part of his life to studying Rivercane and now works with the eastern band Cherokee to try and bring it back.
Chattooga river conservancy video on Rivercane, haven't watched the whole thing myself but it looks really good and detailed
These videos barely have any views or comments, but y'all can help! We can spread the knowledge.
For privacy reasons, I share details online of my real world activities only reluctantly, and not very often. But don't be bamboozled into thinking I have forgotten the Canebrakes. It's exactly the opposite.
I have done a lot of networking and made a lot of contacts. I am not alone. There are other people with a story exactly like mine: first, they heard an offhanded mention of forests of American bamboo, which shattered everything they thought they knew about their environment. Next, they became crazed with fascination, searching for knowledge with insane ferocity. Then, they realized that river cane is not only a plant, it is a keystone species symbiotic with indigenous cultures for thousands of years, and it was almost destroyed due to the subjugation of its habitat and the genocide of its caretakers.
The canebrakes' devotees have been working tirelessly to compile every single scrap of information on canebrakes that exists in writing. Every record, every primary source, every historical mention, every comment and conjecture. I have been given access to some of this priceless treasure trove. The wealth of information is amazing, but even more amazing is how much is still unknown.
The history, properties, and ecological importance of the canebrakes is so much more than I imagined.
For example, the massive amounts of seeds produced by huge canebrakes in flowering events fed the passenger pigeon flocks. Likewise the Carolina parakeet was also dependent on canebrakes, and the extinct Bachman's warbler was a canebrake specialist. The destruction of canebrakes could be responsible for why these birds went extinct.
Canebrakes were absolutely fundamental to the indigenous peoples of the Southeast, providing for their every need. Food, shelter, containers, tools, music and art. The settlers foolishly thought the indigenous peoples were not "advanced" enough for metal tools, but in truth, they already had a material superior to metal. River cane by weight is stronger than steel. You can make knives and blades out of it.
I am excited for the future. It seems like momentum is building to save the river cane and bring back the canebrakes, and I am hoping to join together with all the other like-minded people to accomplish this task.
A new organization has just started in Alabama to bring back the river cane. Here is a blog post to read from a few months ago.
It is not rivercane unfortunately :( Multiple branches off the same node, the branching angle is too wide, and consecutive nodes have branches on the same side
I'm sure OP is already aware, but SECAS has finally been able to add rivercane to their Southeast Conservation Blueprint! Link is to their blog post explaining the addition and what it means for rivercane conservation efforts in the future.
parker leverage 100% would identify as nonbinary. like one day the team is helping a queer youth clinic and one of the kids they interact with uses they/them pronouns. and nate is having a really hard time understanding it, so hardison makes a slideshow of different trans identities. and when he explains non-binary parker’s like “but that’s how everybody feels. no one cares whether they’re a girl or not. it’s just your meat suit.” and suddenly everyone is like ummm parker no we all very much like the genders we were assigned at birth. and parker just. does not believe them.
episode culminates in them saving the clinic and hardison gently broaches that he would still be just as in love with parker if she’s nonbinary. and parker goes “i just don’t see what the big deal is. i’m a thief.” and no one says anything because yeah. parker’s gender is thief.
Happy native plant month! You can use this tool, developed at UMass Amherst, to choose climate-smart native plants for your garden (and find places to buy them) - climatesmartnativeplants.org/plant-selection
Living in Australia, I'm kinda always at the mercy of my posts lining up with crappy timezones, so when I came back to Tumblr and realised there was a fully in-built post scheduling system, I was very excited...only to happily realise that even with that in mind, Tumblr has a much more 'time-blind' and relaxing approach to posts.
It lifts my spirits whenever I see that somebody's gone through my catalogue or liked a post that had been put out days or even weeks ago, and it appears that that's a far more prevalent phenomenon on this site compared to anywhere else. A guy could get used to this more easygoing posting experience, although of course that's all thanks to folks like you! Thank you all.
BONUS:
I happen to think this is a) an incredibly fun fact and b) a pretty funny way to convey it, but despite my efforts (reblogging this once a year), I have yet to get “brain regrowing season” to take off.
They’re the smallest mammal if you measure by mass, coming in at under 2g. That’s tiny.
Because they’re so small, their volume to surface area ratio is messed up, and they are constantly in danger of their body temperature dropping. Their very high metabolism is what keeps them alive.
Their brains can shrink by up to 28% over the winter, so that they don’t have to keep it warm and fed during the coldest season.
They don’t even live where it’s super cold — as the name suggests, they’re found around the Mediterranean, as well as in parts of Southeast Asia. But they’re very sensitive to the cold, on account of being tiny.
