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 US is detransitioning prisoners and just announced that ICE can detain trans people on suspicion of visa fraud. The UK just banned all care for trans kids and is now conducting an "evidence review" into HRT for trans adults and now the biggest DIY HRT web resource just went down.
Trans people need to learn solidarity yesterday, we need to be ready to organise together and set aside anything that we need to in order to get our shit sorted out. Cis people need to take it seriously that the fascists are trying to completely extinguish transess from existence
One immediate step you can take if you're tech literate is try to personally back up as much DIY information from the internet as possible. We need to confront the fact that the internet is American dominated and primarily American-owned and in a larger sense owned and operated by capital primarily. For the informational resources of the internet to benefit human beings we have to deliberately make choices to structure it that way, and in the age of Palantir that largely means thinking about parallel informational infrastructure, setting up another internet or at least other communication systems.
If we want to preserve life saving information, information that aligns with reality over fascism information that is useful, beautiful and true to human beings and ugly and inconvenient to capital, first we have to make backups
tumblr staff will not contact you through anything other than email or their official accounts, which will all have this badge:
DO NOT ENGAGE WITH THIS OR SIMILAR ACCOUNTS AND ABSOLUTELY DO NOT CLICK ANY LINKS FROM IT.
report and block. i'd also appreciate it if you shared this post, bc that blog was JUST created and was already tagging a LOT of people, and i know not everyone has the scam-sensing instinct, even if this might seem obvious to some.
my prized possession is this loveseat I bought from a divorced dad who couldn’t tell me anything about it and in the years i’ve owned it i’ve never been able to find out who made it or where it came from. it’s got nails and finger creases and palm lines but they’re all kinda hard to see in this pic.
finding a new doctor. applying for jobs. searching for apartments. messaging used car dealers. getting your health insurance to do their job. getting a pharmacy to do their job. getting the dmv to accept the documents they told you to bring. just listing things they probably make you do in hell
it's pretty easy to imagine that you are one of some fractional holdout against AI while everyone else has fallen into some misguided love affair with LLMs, and I am so happy to tell you that this is not the case.
the US public is deeply suspicious of AI's impacts on jobs and education. Kamala Harris and the Republican party are both polling better than AI. 8/10 gen zers are concerned about AI's impact on education and only 18% are positive about this technology. there is widespread, bipartisan grassroots organizing against data centers. 97% of Britons are against Grok's "undressing" technology. the majority of Americans are concerned about AI in arenas like self-driving vehicles and healthcare. Even polling data from companies centered on AI shows significant concern around generative technology. OpenAI isn't meeting internal growth bench posts. On top of all that, Musk and Altman are currently both making fools of themselves in a very public trial.
its 2026 i cannot handle any more fucking "author A obviously ripped off author B" discourse by people Who Have Only Seen the work of author B and admit themselves that they have no further knowledge of the literary landscape they are moving in. like.
I’ve read that Agatha said she loved being married to an archaeologist because the older she got, the more interesting he found her. And I think that is one of the best quotes about love that I have ever heard.
Google is transforming Search from a list of links into an AI-powered experience filled with conversational answers, autonomous agents, and
The era of the “ten blue links” is officially over.
At its Google I/O conference on Tuesday, Google unveiled an AI-powered overhaul of Search centered around a reimagined “intelligent search box” — what the company describes as the biggest change to this entry point to the web since the search box debuted more than 25 years ago.
Instead of returning a simple list of links, Google Search will drop users into AI-powered interactive experiences at times. Google is also introducing tools that can dispatch “information agents” to gather information on a user’s behalf, along with tools that let users build personalized mini apps tailored to their needs.
The resulting experience will no longer look much like how people envision Google Search, which has long been defined by ranked links to websites that have the information you need.
With the revamped Search experience, the new search box simply expands to accommodate longer, more conversational queries, rather than making you decide what type of search experience or mode you want to choose at the start of your query. It will also have a new AI-powered query suggestion system that goes beyond autocomplete to help people craft more complex and nuanced queries, Google says.
Google’s AI Overviews will also allow users to ask follow-up questions in AI Mode, beginning Tuesday, the company noted.
"Combined, these changes will likely further decimate Google referrals to publishers, which have already been suffering from declining referrals due to AI Overviews. This has put some ad-dependent media operations out of business, and now things will likely get worse."
I think you can tell a lot about how rigorous and committed someone's belief in a human right is by how quickly they are able to name people who they think could or should have that right taken away.
It's wild to me that people in the notes are all arguing about rapists and murderers and people deserving to get their human rights taken away for doing Bad Things, and...
Yeah, sure, serial killers deserve human rights too, but isn't there a more obvious demographic you're sliding right by? Isn't there a demographic of people so thoroughly erased by human rights discourse that their rights aren't even debated, it's just taken for granted that human rights don't apply to these people?
(It's minors. I'm talking about minors. Also disabled/neurodivergent adults under institutionalization/guardianship who have been reduced to the legal status of minors.)
I literally do this as a first-day activity in my childhood studies courses.
I take a poll: "how many of you would agree that 'everyone deserves the right to privacy' is a pretty uncontroversial statement?"
when 95% of them have put their hands up, I say "now, what if I clarify that 'everyone' includes children?"
and as everyone lowers their hands slowly and gives me a confused look like a deer in headlights, I tell them "okay. this class is about what it means to not be part of 'everyone'."
the party of law and order who is tonsil deep on the cock and balls of palantir and flock who claims every foreign born person on US soil commits 100 crimes a day and should be put in a concentration camp for jaywalking is very upset about being surveilled and ticketed while committing crimes on a foreign nation's land. because as we know if the government owns the cameras that's communism but if a private company ran by a pedophilic vampire billionaire owns the cameras, that's actually democracy.
2087
dailybridgerton
jbbartram-illu
pixelskylines
prguitarman
baratheonpilled
discoholicmusic
