"There's a whole world out there waiting for you."

Lau. Canadian. She/Her. I like to write stuff sometimes. Header image credit to @vvivaa

[emmaharwoods]

you can click on this button once daily to help palestine and support other causes in the middle east for free. it takes literally 5 seconds and could help save lives so please take the time to click and share this link.

[grelliam]

#oh it raises money thru ad revenue and ur click confirms the value to advertisers. neat

[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.

[irispurpurea]

This is the paper. It's excellent, highly recommend reading it.

I remember reading about Gebru's firing but I had no idea this was the paper she was fired over.

[depsidase]

[dearest-lady-disdain]

bold words for a man who has been killing people for eleven episodes straight

[homunculus-argument]

Fascinated by the people who comment on online recipes like

"I tried making this vegetable dip! I didn't have mayo so I just used low fat yogurt instead, and we are doing keto so I replaced the sour cream with lard. My husband is deathly allergic to citrus fruit so to get the same tang I added vinegar instead of lime juice. I used pancake mix instead of garlic, and since I didn't have salt, pepper, paprika, onion powder, parsley, or dill, I substituted with pumpkin spice and ground-up civet skull. My husband hated it :( I don't know what I did wrong."

Well ma'am, first of all, you made the Dip of Theseus.

[foenixed]

"Why would this character be jealous of their friend's romantic relationships? Clearly they're in love with them" I don't know if you guys have ever Had Friends but it's actually pretty common for people to feel jealous if their best friend suddenly has less time for them and is prioritising someone they just met over them. That's not inherently romantic, that's the general human desire to be valued by the people you value.

[kiragecko]

My Very Little Sister had to move out of 3 living situations because her friends/cousins found romantic partners and didn't need a roommate anymore. She started getting really grouchy whenever a friend mentioned romance after a while.

My cousin lost his best friend of 20+ years because the friend's girlfriend decided he shouldn't see anyone else. For years my cousin was leery about anyone else starting romantic relationships, and acted kinda jealous and resentful.

I had to make clear boundaries with my now-husband, explaining that I was NOT going to be investing all of my limited emotional energy into our relationship, he had to be willing to share. And my emotional energy is VERY limited, so he struggled! If I hadn't been so very very aromantic, and had spent as much time with him as he wanted, my other relationships would have suffered greatly. As it was, people had to get used to a changed dynamic with me - he came everywhere, because it was easier to get out of the house when he was dragging me.

People in platonic relationships have very real reasons to be jealous!

[divorcedwife]

studying history is like. here's to another beautiful day of not being pregnant and of having no obligation to ever be. thank you women who fight for abortion and contraception and independance from men for another beautiful day of not being pregnant and of having no obligation to ever be

[roycohn]

stop saying "gen z brought back bush-era purity politics" i grew up in the bush era and even then people weren't saying that you're a sex addict for having boring marital sexual congress in the same house as your children. this is just plain unhinged

[prof-grace-aspen]

Literally almost every millennial I know has a memory of accidentally walking in on their parents or hearing their parents having sex. It's fucking normal. Human beings have sex. Your parents fuck. Get over it. Being weird about it isn't healthy.

[whatsnewalycat]

I really loved Robert Evans’s response to this

[wafflewarrior111]

So... I found this and now it keeps coming to mind. You hear about "life-changing writing advice" all the time and usually its really not—but honestly this is it man.

I'm going to try it.

[holyfunnyhistoryherring]

[Image description: paragraphs of black text on a white background.

Text in the image reads:

Writing advice by Chuck Palahniuk. In six seconds, you'll hate me. But in six months, you'll be better writer.

From this point forward — at least for the next half year — you may not use "thought" verbs. These include: Thinks, Knows, Understands, Realizes, Believes, Wants, Remembers, Imagines, Desires, and a hundred others you love to use.The list should also include: Loves and Hates. And it should include: Is and Has, but we'll get to those later.

Until some time around Christmas, you can't write: "Kenny wondered if Monica didn't like him going out at night..."

Instead, you'll have to unpack that to something like: "The mornings after Kenny had stayed out, beyond the last bus, until he'd had to bum a ride or pay for a cab and got home to find Monica faking sleep, faking because she never slept that quiet, those mornings, she'd only put her own cup of coffee in the microwave. Never his."

Instead of characters knowing anything, you must now present the details that allow the reader to know them. Instead of a character wanting something, you must now describe the thing so that the reader wants it.

Instead of saying: "Adam knew Gwen liked him." You'll have to say: "Between classes, Gwen had always leaned on his locker when he'd go to open it. She'd roll her eyes and shove off with one foot, leaving a black-heel mark on the painted metal, but she also left the smell of her perfume. The combination lock would still be warm from her butt. And the next break, Gwen would be leaned there, again."

In short, no more short-cuts. Only specific sensory detail: action, smell, taste, sound, and feeling.

Typically, writers use these "thought" verbs at the beginning of a paragraph (In this form, you can call them "Thesis Statements" and I'll rail against those, later). In a way they state the intentions of the paragraph. And what follows, illustrates them.

For example: "Brenda knew she'd never make the deadline, was backed up from the bridge, past the first eight or nine exits. Her cellphone battery was dead. At home, the dogs would need to go out, or there would be a mess to clean up. Plus, she'd promised to water the plants for her neighbor..."

Do you see how the opening "thesis statement" steals the thunder of what follows? Don't do it.

If nothing else, cut the opening sentence and place it after all the others. Better yet, transplant it and change it to: Brenda would never make the deadline.

Thinking is abstract. Knowing and believing are intangible. Your story will always be stronger if you just show the physical actions and details of your characters and allow your reader to do the thinking and knowing. And loving and hating.

Don't tell your reader: "Lisa hated Tom."

Instead, make your case like a layer in court, detail by detail.

Present each piece of evidence. For example: "During role call, in the breath after the teacher said Tom's name, in that moment before he could answer, right then, Lisa would whisper-shout 'Butt Wipe', just as Tom was saying 'Here'."

One of the most common mistakes that beginning writers make is leaving their characters alone. Writing, you may be alone. Reading, your audience may be alone. But your character should spend very, very little time alone. Because a solitary character starts thinking or worrying or wondering.

For example: "Waiting for the bus, Mark started to worry about how long the trip would take..."

A better break-down might be: "The schedule said the bus would come by at noon, but Mark's watch said it was already 11:57. You could see all the way down the road, as far as the Mall, and not see a bus. No doubt, the driver was parked at the turn-around, the far end of the line, taking a nap. The driver was kicked back, asleep, and Mark was going to be late. Or worse, the driver was drinking and he'd pull up drunk and charge Mark seventy-five cents for death in a fiery traffic accident..."

A character alone must lapse into fantasy or memory, but even then you cannot use "thought" verbs or any of their abstract relatives.

Oh, and you can just forget about using the verbs forget and remember.

No more transitions such as: "Wanda remembered how Nelson used to brush her hair."

Instead: "Back in their sophomore year, Nelson used to brush her hair with smooth, long strokes of his hand."

Again, unpack. Don't take short-cuts.

Better yet, get your character with another character, fast. Get them together and get the action started. Let their actions and words show their thoughts. You — stay out of their heads.

And while you're avoiding "thought" verbs, be very wary of using the bland verbs "is" and "have".

For example: "Anne eyes are blue."

"Ann has blue eyes."

Versus: "Ann coughed and waved one hand past her face, clearing the cigarette smoke from her eyes, blue eyes, before she smiled..."

Instead of bland "is" and "has" statements, try burying your details of what a character has or is, in actions or gestures. At its most basic, this is showing your story instead of telling it.

And forever after, once you've learned to Unpack your characters, you'll hate the lazy writer who settles for: "Jim sat beside the telephone, wondering why Amanda didn't call."

Please. For now, hate me all you want, but don't use thought verbs. After Christmas, go crazy, but I'd bet money you won't.

(...)

For this month's homework, pick through your writing and circle every "thought" verb. Then, find some way to eliminate it. Kill it by Unpacking it.

Then, pick through some published fiction and do the same thing. Be ruthless.

"Marty imagined fish, jumping in the moonlight..."

"Nancy recalled the way the wine tasted..."

"Larry knew he was a dead man..."

Find them. After that, find a way to rewrite them. Make them stronger.

/End text. End description.]

[guooey]

Love him... love mr. cool...

[existential-dread-in-the-am]

So deer have this instinct to stand tall and walk like a king when they notice something (a predator) watching them so they look like strong opponents since predators usually target the weakest members of the herd. The baby is doing the Don't Eat Me Walk.

[despicableplankton]

Predators: “How could I possibly eat someone THAT cool??”

[fibromyalgicaf]

Edited to add: Since a lot of people are reblogging this original post, I'm adding the updated version I did that incorporates the intersex circle...

I know intersex people are still getting excluded in a lot of LGBTQIA+ spaces (let alone wider society) and I think it's crucial to show this group is included in the statement that we all deserve equal rights.

[galliaestomnis]

Underwater environments transform into phantasmagorical landscapes through the lens of photographer Nicolas Floc'h

More of his outstanding work here : www.nicolasfloch.net

[simplysnaps]

Hannah Montana is fucked up because its entire POINT as a show is that children should be protected from fame and exploitation, but it stars a REAL little girl that's being exploited. Nearly every episode carries the looming threat of Miley being outed as Hannah and losing her peaceful teenage life to the ravages of fame. Her father in the show (played by her own father in real life) wisely protected her from the trauma of fame by making her wear a disguise and live a rather quiet, interview-free life. Meanwhile the REAL Billy Ray Cyrus sold his daughter to Disney Channel when she was 11 and forced her to read dialogue about how terrible it would be to face the public eye. Like... Jesus, dude. The fictional Robby Ray is 10x the father, and it's not even close. (It's also IMMENSELY funny that her dad doesn't use his real name in the show, while she does. Almost like he wanted a bit of a disconnect between his identity and his character. Something Miley didn't get.)

[annabelle--cane]

you simply have to make your own post. there's no choice about it you Have to learn how to Make Your Own Post. I'm begging and screaming for you to make your own brand new original post and not say That on My Post. it's so easy. it might even be cool and mature and darkly chic and secretly leftist to make your own post. making your own post is like wearing headphones to watch tiktoks on public transit, probably, in a way. river water, socks, your own post, etc. please.

[riverlarking]

i love clicking on somebody’s ao3 profile and seeing the most nonsensical collection of fandoms. like yess let's live a thousand lifetimes

[heycallmered]

people will really come into kink spaces and say you can't forcefem women like there wasn't a feature length movie about an elderly gay man forcefemming a woman as part of scheme to thwart an elaborate assassination attempt before the killer even determined their target

[llama2237]

What... What movie is this.

[heycallmered]

[heycallmered]

ain't no way in hell this post even breaks 500

[marmoset-marmoset]

i was trying so hard to remember the nonexistent assassination subplot in My Fair Lady

[hydrachea]

> turns on my computer

> disables a new AI feature that was turned on by default

> opens my email

> disables a new AI feature that was turned on by default

> launches a software

> disables a new AI fea