De interés

Donde vendrá a parar todo lo que encuentre útil e interesante para compartir.

[lasnotasdeleo]

Publicaciones de interés

Para mayor comodidad, he aquí las divisiones que utilizo:

Estudios: Análisis de personajes, tramas y temas de distintas series.

Recursos: Todas aquellas publicaciones que contengan sitios o consejos para la escritura y la ilustración.

Referencias: publicaciones con detalles que me gustaría inlcuir en mis ilustraciones o escritos.

[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.

[1000-life-hacks]

[archeologist-prophet]

Lovely to see we have spaces where you can gain access to so much literature!

[midnights-dragon]

In case anyone is having a bad night

(The best of this post and its reblogs, but with links that work)

Here is a website where you can scroll down to all the different levels of the ocean 

Here is a website where you can see the future of the universe

Here is a website where you can press a ‘make everything okay’ button, over and over, until things really are okay

Here is a website that you can read if you feel like a burden

Here is a website where you can look at strobe illusions (TW strobe/flashing)

Here is a website where you can cut stuff up (TW blood/sh)

Here and here are websites where you can play with sand

Here is a website where you can draw with macaroni and other fun foods

Here is a website where you can paint someone’s nails

Here is a website where you can grow a garden with emojis

Here is a website with hundreds of videos of people hugging you (rightfully dubbed ‘the nicest place on the internet’ because it really is, y’all, it made me cry)

Here is a website that will take you to other useless websites

Here is a website where you can make a tiny cat play bongo drums (and other instruments!)

Here is a website to help give you gentle reminders <3

Here is a website where you can grow a tiny farm

Here is a website where you can take a bunch of scientific personality tests

Here is a website of calm rain noise

Take a breath. It’s going to be okay, I promise.

[midwinterhunt]

hot artists don't gatekeep

I've been resource gathering for YEARS so now I am going to share my dragons hoard

Floorplanner. Design and furnish a house for you to use for having a consistent background in your comic or anything! Free, you need an account, easy to use, and you can save multiple houses.

Comparing Heights. Input the heights of characters to see what the different is between them. Great for keeping consistency. Free.

Magma. Draw online with friends in real time. Great for practice or hanging out. Free, paid plan available, account preferred.

Smithsonian Open Access. Loads of free images. Free.

SketchDaily. Lots of pose references, massive library, is set on a timer so you can practice quick figure drawing. Free.

SculptGL. A sculpting tool which I am yet to master, but you should be able to make whatever 3d object you like with it. free.

Pexels. Free stock images. And the search engine is actually pretty good at pulling up what you want.

Figurosity. Great pose references, diverse body types, lots of "how to draw" videos directly on the site, the models are 3d and you can rotate the angle, but you can't make custom poses or edit body proportions. Free, account option, paid plans available.

Line of Action. More drawing references, this one also has a focus on expressions, hands/feet, animals, landscapes. Free.

Animal Photo. You pose a 3d skull model and select an animal species, and they give you a bunch of photo references for that animal at that angle. Super handy. Free.

Height Weight Chart. You ever see an OC listed as having a certain weight but then they look Wildly different than the number suggests? Well here's a site to avoid that! It shows real people at different weights and heights to give you a better idea of what these abstract numbers all look like. Free to use.

[narrettwist]

Homie gonna share this

[artifacts-and-arthropods]

Planthopper Parasite Moths: the caterpillars of this family are parasites that attach themselves to the bodies of other insects and then gradually feed on the fluids within the host's abdomen

Above: the larval form of Fulgoraecia exigua

Caterpillars of the family Epipyropidae, commonly known as planthopper parasite moths, are ectoparasites that feed on the haemolymph (i.e. "blood") of planthoppers and cicadas. This family contains at least 40 described species, all of which are parasites or parasitoids.

Above: planthopper parasite caterpillars tucked beneath the wings of their hosts

This is one of the few known examples of a caterpillar that actually engages in parasitic, carnivorous and/or predatory behavior.

Above: Fulgoraecia exigua

The caterpillar has hooked claws that allow it to cling to the host's body; it uses its mandibles to penetrate the cuticle around the abdomen, then inserts a proboscis-like structure into the planthopper's body and begins to feed on the fluids within.

Each caterpillar spends about 4-6 weeks feeding on its host, with its body often tucked beneath the larger insect's wings.

Above: a Fulgoraecia exigua caterpillar can be seen feeding beneath the wings of this planthopper

Young planthopper nymphs that are parasitized by these caterpillars rarely survive the process. The survival rate for older nymphs and adult hemipterans is also quite low.

Above: the photo at the top shows an Epipomponia nawai caterpillar feeding on a cicada, while the photo on the bottom shows Fulgoraecia exigua feeding on a leafhopper

Planthopper parasite caterpillars are often tiny, with their bodies measuring just 3-7mm long, and they're covered in waxy white filaments that make them look like little cottonballs. These features seem to mimic the "fluffy" appearance of many fulgoroid planthopper nymphs, which may enable the caterpillar to sneak up on its host.

Above: planthopper parasite caterpillars

When the caterpillar reaches maturity, it finally detaches from its host and then uses a thin strand of silk to abseil down to a leaf or a branch, where it can spin a cocoon around its body and enter pupation.

Above: the cocoons of Fulgoraecia exigua

The caterpillars of this family create very distinctive cocoons. As delicate layers of silk are folded together around the pupal case, they often form ridges, cone-like structures, or wider, flatter folds that look almost like rose petals.

Above: a cocoon made by an unidentified moth from family Epipyropidae

The fully-developed moth may emerge weeks or even months later, depending on the species. The adults of this family typically have a blackish-gray appearance.

Above: the adult form of Epipomponia nawai

Other examples of predatory, carnivorous and/or parasitic moths can be found in my earlier post about meat-eating caterpillars.

Sources & More Info:

Journal of the Lepidopterist's Society: Predatory and Parasitic Lepidoptera

The Lepidoptera: Form, Function, and Diversity: Epipyropidae

Moths of North Carolina: Fulgoraecia exigua

Bug Guide: Family Epipyropidae

Bombay Natural History Society: The Biology and Morphology of Epipyrops eurybrachydis

Egyptian Journal of Biological Pest Control: Parasitism by Fulgoraecia melanoleuca

Journal of Asia-Pacific Entomology: Behavioural and Phylogeographic Observations on Epipomponia nawai

Species Connect: Carnivorous Butterflies and Moths

[carmahnart]

Hey y’all I have an announcement! My web app that I’ve been working on, Afro Index, is now live! It’s a visual reference library of Black hairstyles, for artist, animators, writers, and anyone who wants to learn more about them!

Check it out at afroindex.org! 💛✨

Afro Index – Black Hairstyle Reference Library

A reference library for Black hairstyles with accurate naming, structured filtering, and curated reference images.

[redstonedust]

pet peeve is when you look up fashion references from a specific era and you keep getting modern day '[era]-inspired' fashion like NO i want authenticity damn it. i can see your 2020 photo quality and your 2020 hair and your 2020 makeup. youre not fooling me.

[bincliff]

hello i'm a historical fashion researcher and i have a lot of experience looking up things! this is a very widely experienced irritation and you're definitely not alone in this, but i am here to share everything i know!

so, ways to get around this:

turn off AI results. they're literally nonsense to us

don't use pinterest because the sources/provenance is often hard to trace

a standard internet search can be okay, but museum collections are the top tier (list of collections below this list)

instead of broad terms like victorian, regency, tudor, renaissance etc. try using the decade you're looking for. if you're not sure of what decade it is but have a vague image in your head, look on the fashion history timeline and just jump around until you find it. but even changing to e.g. 19th century will give better results than victorian

including terms like womenswear/menswear, daywear, formal wear, evening wear, court dress should increase the value of your search too

including "fashion plates" in your search can give you a nice impression of the intended silhouettes of the era. some of these might be a little stylised but will show you what was considered in vogue

for pre-fashion plate eras or things like makeup and styling, you'll have to look at portraiture or manuscripts. these are harder to actually find what you're looking for, but searching museum collections and limiting results to specific date ranges will be your friend

when looking at art, do bear in mind sometimes artists would paint fabric extra flow-y to show off their skills. it might not have been exactly like that in terms of fabric weight or drape. so, a pinch of salt required!

if you find something on image search where the provenance is dubious, reverse image search and you might find a source! i've been able to trace random pinterest images to real sources, but this does take a lot of time and effort and is often not worth the headache

some online resources and museum collections:

fashion history timeline is an invaluable resource if you're trying to get a feel for everything and should be your first port of call. it'll also link to good examples

the met has a vast number of extant examples of clothing, as well as fashion plates

costume institute fashion plates is a subcollection of the met for fashion plates (1800s-1922)

v&a also has many extant garments, fashion plates, and incredible articles on clothing and aesthetics. read the details of the objects because they'll often reveal a lot about the piece

lacma is good for C19th-20th pieces

nypl digital collection for photographs

national portrait gallery or similar for portraiture, or literally any museum in your country that has historical art

national museums scotland can be useful situationally but might be oddly specific

stout style history is a great collection for finding image references for fat people wearing historical clothes. survival bias of a lot of museum pieces tends towards smaller clothing that couldn't be repurposed, but this aims to counter that. it's not sortable, but is still a really nice resource

wikimedia commons is surprisingly handy! and the images, if you should need to link/repost them, are public domain

auction websites sound like a funny one to recommend. some won't have mannequins and some will. just look up historical garment auctions and you'll find some!

anyway, i hope this has been a good place to start for anyone interested! there are probably some i've missed because there are so many museums across the world and i don't know about all of them or can't remember them. but these are the ones i've used the most! (my specialisation/jobs i've had to research for have only really been in western fashion, so my resources reflect that)

[haemey]

Wikipedia has a list of fashion museums. Unfortunately, the page itself is only available in German, but the introductory paragraph is very short and after that, it's organised by country, and then it's a simple list. If you click on a museum's article, the website is usually linked in the overview table.

[wild-e-eep]

Mystery Webs!

I come across these webbed-up fungi occasionally. They look quite like the tent-webs of some moth and sawfly larvae, but differ in occurring later in the year; enclosing a fungus instead of a food-plant; not being full of specks of caterpillar dung; you also rarely see any living (or dead) creatures, eggs or pupae in them.

The cleanness of the webs and the way they are restricted to a particular fungus might make you think you are looking at a grey, fuzzy mould.

To find out what was making these, I took a piece home and kept it in a vivarium. Checking on the vivarium in the middle of the night I spotted a small mealworm-like creature moving through the web, leaving a slime trail behind it. I caught it and put it in an empty enclosure where it soon started fervently weaving by bobbing its head back and forth at an impressive rate, extruding silk from spinnerets on its snout. In the morning, the new enclosure was full of webs, and the larva resting in the middle.

I believe these larvae are fungus gnats of the family Mycetophilidae - a large family of small, rather plain flies with thousands of described species, most of which lay their eggs on fungi. I haven't found much witten about the function of the webs: maintaining a humid atmosphere; excluding other small creatures; catching spores for consumption have all been mooted. The larvae also seem to be able to move smoothly and rapidly through their silk network.

[westformiles]

@onenicebugperday

[lasnotasdeleo]

Awesome! It looks so pretty!

[spinji]

At this point if it's not canon that Val is an abused sex worker that's continuing the cycle by hoarding what little power he can get and using it against whoever he can force below himself because he's terrified of being weak again I'm going to be so fucking confused.

It explains why he clings to Vox so much, him having a man in power that also let him be in power as an equal. Someone who's safe without infantilizing or controlling him (Val being immune to hypnotism).

It explains why he acts the way he does with Angel, emphasizing control over his personhood and autonomy. If he's the abuser he cannot be the abused and the most upset he gets is when he's losing a significant amount of his control over what Angel does (Angel moving out).

It explains why he sexualizes himself just as much as he sexualizes others. It's a lifestyle that's always been a part of him and taking control of that has been a great way to reclaim himself after what he went through.

It sets up numerous ways for him to have fallen to sin, whether it was lashing out at his abusers like Angel did or resorting to things like stealing from his clients to get ahead.

It sets up the perfect cycle of abuse story between him and Angel, Val symbolizing a person Angel might become if he regresses his progress and serving as a manifestation of every element of his character that led him to ending up in Hell. Rejecting that, being better than that, will be what let's Angel ascend like Pentious did; confronting and changing the part of himself that led him to sin.

Hell- it explains why he's a moth specifically. "Moth" is a Victorian era slang term for a prostitute.

I hate when fandoms lean on popular fanon expectations for a series then get genuinely mad when the showrunners do something different. There's a lot of theories and expectations I can let go of to see the series' actual vision but I will actually be pissed if none of this is true.

[yet-the-world-spins-on]

Just caught up with Spy x Family.

I don't quite understand why this is a hot take, but, uh, Loid definitely knows that he loves Yor and Anya. (Now, to be fair, he almost certainly doesn't realize how much Yor likes him. He is genuinely bad at reading the women in his life, which is partly bc because they're actually not great at expressing themselves, and partly because he assumes any time something goes wrong it's because he messed up)

The thing is, Loid's not willing to act on the love he has for them to turn the mission-family into a real-family. Which is partly bc of his personal experiences, yes! It's also partly bc he's still going to be a spy. How can he commit himself fully to Yor and Anya when he's still working for WISE? He believes he can't (and I suspect he's right), and he believes that the work he does with WISE is too important to walk away from.

I genuinely think he believes that loving them means not "turning this into a real family" -- because if he makes the marriage official, then Yor thinks the family is real and Twilight knows it still isn't. Which is making me think Athelas-and-Troika-City-Between thoughts, but that's for another time. (To those who haven't watched Spy x Family: Yor and Loid have entered a marriage of societal convenience; there is no sexual aspect to it).

Now, is he deliberately not thinking about how much he likes Yor and would like to genuinely marry her? Probably. But he's not avoiding the thought out of, what's the word, out of "oh that's too scary to think about," he's avoiding the thought because he can't act on it. There's no point fantasizing.

Which creates a really interesting foil between Twilight and Nightfall, bc Nightfall is spending all her time fantasizing, even though she can't act on it. And we see what that is doing to her -- we see that the fantasizing is destroying her, even though Twilight doesn't know about it. You know how people say "well this is private, it's only affecting me, why do you care?" This is why.

[afeelgoodblog]

Why hasn’t this been done before?

[autisticexpression]

You know why.

[petrosapian]

Mind the Gap: a handbook of clinical signs on black and brown skin

Second year medic, Malone Mukwende, has been working with staff members as part of a student-staff partnership project looking at clinical t

[missmixx-a-lot]

Cause racial health disparities…

[xmagnet-o]

I hope this gets published

[algorizmi]

It is available for download as a pdf from their website. www.blackandbrownskin.co.uk/mindthegap

[imperatorsapphiosa]

Hey, if anyone’s curious, this IS making a serious impact — this handbook was taught as standard practice for my EMT license course, as well as several other programs. A lot of new providers are being taught this as a matter of course.

[indeedgoodman]

Basket Star.

[marine-gifs]

A persian carpet flatworm. From The Mating Game (2022).

[your-local-house-cricket]

uses him to scroll on tumblr for 2 hours

[hamletisintown]

I was baffled so I had to look it up but yeah, it IS in fact a real fucking beetle (possibly Hololepta aequalis just by looking at them but I could very well be wrong bc i don't know shit about this family)

Hololepta - Wikipedia

[rivriderart]

I was asked by a friend yesterday if I could offer basic tips about comic paneling. As it turns out, I have a lot to say on the matter! I tried breaking down the art of paneling using the principles of art and design, and I hope it helps you out!

EDIT: uh uh there are a lot of people reblogging this, so i figure i may as well append this now while i can lol

This whole thing was very much cranked out in a few hours so I had a visual to talk about with a friend! If this gives you a base understanding of paneling, that's awesome! Continue to pull in studies from the comics you see and what other artists do well and don't do well! You can tell paneling is doing well when the action is flowing around in its intended reading format.

Here's the link to the globalcomix article from which I pulled the images about panel staggering! Someone sent in a reblog that it wasn't totally clear that the 7th slide mostly covers what NOT to do in regards to staggering, and that is my mistake!

I saw in a tag that someone was surprised I used MamaYuyu too, and I don't blame them lol. If I had given myself more than a couple hours maybe I would have added something else on, I just really admire MamaYuyu's paneling personally.

uh uh, final append: I am by no means a renowned master of paneling, so if you find anything off base here, by all means, counter it with your own knowledge and ways you can build upon from here! Art is always a sum knowledge of everything we find. 💪

[lilkakow]

Images created solely using mathematical equations by mathematical artist Hamid Naderi Yaganeh