wear a fucking mask

30-40, they/them, white, nonbinary, anarchist. I don't tag anything so please don't follow if that's a problem, sorry. 100% reblogs, oc free. i have a sideblog of just cute animals being cute called @sereminarshappyplace icon by @casmarotta

[silentevilx-deactivated20160627]

SO I FIND OUT MY FRIENDS AND I ARE ON OUR LOCAL NEWS WEBSITE

We were asked all these questions about how long the costumes took us, how we did it, etc. 

I paused the video right when we were asked how much money we spent on our cosplays and

fjdkaf;jkd;al

does that give you your answer

[manafromheaven]

THIS WILL NEVER NOT BE FUNNY 

[no-one-offical]

‎

[keyboard-worrier]

If you're ever struggling with math, Fields Medalist Yu Deng recommends yuri.

[incoherent-orca]

god bless him and may his egg crack soon

[circlusion]

The world has irrevocably changed 😃

[so-much-for-subtlety]

There’s also already five cases of cops stalking their wives, mistresses, ex-girlfriends, ex-girlfriends new boyfriend etc like this - there no warrant required, no supervision around who or when cops can access these systems.

Police departments don’t even have systems in place for when cops retire or fired to make sure their Flock credentials are revoked. Crazy days.

[t4timekeeper]

ACCIDENTALLY LEFT MY MACERATED APPLE SYRUP ON THE STOVE AND IT CARAMELIZED I GUESS?

[t4timekeeper]

it was super thick and dark and golden and bubbly so i took it off the heat. happy accident.

[t4timekeeper]

tastes delicious so good. idk what to do with it though

[t4timekeeper]

thank you for your recommendation! with brief consideration, it has been disregarded.

[can-i-make-image-descriptions]

Plain text: accidentally left my macerated apple syrup on the stove and it caramelized I guess?

[Image #1: Tumblr reply from evil-tboys reading: Is this good or bad

Image #2: Tumblr reply from evil-tboys reading: Spread it arpund (around) the carpet and give yourself an ant infestation (thumps up emoji) /End IDs]

[dankmemeuniversity]

[verminfang]

Thinking about a vampire dude bro who won't drink girl blood- because he's afraid it will estroginise him soy products style- getting trapped in a mental cage when he learns vampire dude who only bites other dudes reads as gay

[islamprotestan-blog]

only 5 seconds

[zanyflowerpizzatrash]

i wasn't ready for that..lol

[suffer-no-humiliation]

Immediately recognized that look

[unheardoracle]

Your dialogue is stiff because everyone is being too honest

Great dialogue can come from radical emotional honesty. But in most real conversations, people

dodge questions

answer indirectly

lie (badly)

change the subject

crack jokes to avoid serious answers

lie (convincingly)

say half of what they mean

tell the truth but get so anxious/angry/defensive others now think they're lying

If your dialogue feels unnatural, it might be because everyone is saying exactly what they think, or articulating their thoughts too well. Try letting characters hide things instead.

Subtext is where tension lives.

[dopedyk]

more so,

people don’t always speak in full sentences,

slang can change depending on age range, demographic, etc. &

don’t be scared of vocal pauses (uhm, uh, err)

[dragon-in-a-fez]

just saw a girl wearing two broken smartwatches on the same wrist. some kind of grunge scifi protagonist perhaps

[dragon-in-a-fez]

I'm a little obsessed with you if that's cool

[justalurkr]

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.

[buam]

On the Dangers of Stochastic Parrots: Can Language Models Be Too Big? by Bender et al

[justalurkr]

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.

[kauricat]

This is an excellent read.

I just went looking for the post on X and got a message that it didn't exist

I searched for Guri Singh and got search results showing his account existed, but when I clicked on it I got a message his account did not exist Does anyone know if he made his account private or if he got nuked by Elon? Or do I just suck at X (because I never go there)?

[chaos-of-nyx]

Yup, the tweet is unavailable because the account has been suspended.

It doesn't look like Wayback Machine has a copy, either.

So I've found a few other references:

Paper: On the Dangers of Stochastic Parrots

Presentation of the paper to The Turing Institute

MIT Technology Review: We read the paper that forced Timnit Gebru out of Google. Here’s what it says.

The Guardian Interview: ‘There was all sorts of toxic behaviour’: Timnit Gebru on her sacking by Google, AI’s dangers and big tech’s biases

DAIR Institute

[mcromwell]

Does anyone have that photo of the kid's Blue Angels protest sign? It was like "die, crash blue angels" or something, really funny, cannot find it anywhere online

[mcromwell]

Turns out I had it THE WHOLE TIME! golly

[sapphling]

and honestly it's just fucking insane that every time we try to talk about harm to children in particular we have to talk circles around the fact that these things are very, very overwhelmingly done by parents/family members/teachers/religious leaders/known and trusted adults in a position of power over a child, to the extent that many people's suggested solution to child abuse is "give the people most likely to abuse a child EVEN MORE strangling legal and material power over the kid"

[sapphling]

you're not supposed to actually address who is abusing children and what can be done to give those children recourse and the ability to leave abusive situations because nobody actually gives a shit about child abuse beyond the familial property rights violation it represents to them and the cudgel against minorities which it can be turned into

[shower-thoughts-last-responder]

You see something broadly similar in nursing where you have to tip toe round the fact that nurses can directly and intentionally harm patients; one of the most common phrases you'll hear when trying to address an issue is something like "well no nurse is coming to work deliberately wanting to hurt their patients". I don't say this to derail from the issue of protecting children from their parents/guardians, i say it because i think it highlights the consistent problems that not offending the sensibilities of people in positions of power over vulnerable people is seen as more important then actually protecting those vulnerable people. "How dare you suggest I'm a bad parent, I'm not going to listen to anything you say because clearly you just hate me, and my feelings are clearly the real issue now" type mindset that means you end you talking circles around issues trying not offend anyone instead of directly addressing what is wrong

[whaile]

late summer / early fall thoughts

[dark-miku]

CPU swaps are the most intimate type of repair. you can't do it live, you have to take the system down. gotta pull the whole thing out and remove the cover, use your special torx driver for the heat sink, delicately wipe off the paste... then pull out her cpu, that bad, malfunctioning chip... all those thousands of pins are exposed... she's so vulnerable right now. every move counts. have a good look around in there to check for debris. carefully drop in the new chip, don't touch anything else. apply the thermal paste—you're the only one who will ever know how it looks right now. seal her back up with the heat sink, making several passes around the screws for even pressure. put her lid back on and power her up again, and find out how you did. have fun waiting out the half-hour POST sequence in suspense. you're a neurosurgeon now.

[rackmount-official-my-ass]

pointing this in the direction of @virtualgirladvance for no particular reason

[virtualgirladvance]

some day ill trust someone enough...

[muzzlesbian]

shoving the tardis up my neopussy call that time dilation

[muzzlesbian]

-the doctor telling me that my surgery site is healing well at my 6-month checkup

[stillnaomi]

the fact that many liberals care more about whether a state is "authoritarian" than whether it lifted massive swathes of its population out of poverty says a lot about their politics. they say China is bad because it's authoritarian and brush aside the 800 million people it's lifted out of abject poverty. they complain that Cuba is "authoritarian", while ignoring that it brought healthcare and education to the masses, who were left to suffer under the Batista regime. at some point you have to realize that you're condemning a country that's run in the interests of the working masses of its population

the kicker is that the marxist-leninist countries being criticised are more democratic than the liberal democracies the critics are comparing them too

[omgellendean]

Example from today:

[tamamita]