It's the human condition @irrationalgoods - Tumblr Blog

A summary of A Gathering of Shadows by Victoria Schwab

This AI is bad at drawing but will try anyways.

There was a paper recently where a research team trained a machine learning algorithm (a GAN they called AttnGAN) to generate pictures based on written descriptions. It’s like Visual Chatbot in reverse. When it was just trained to generate pictures of birds, it did pretty well, actually.

(Although the description didn’t specify a beak and so it just… left it out.)

But when they trained the same algorithm on a huge and highly varied dataset, it had a lot more trouble generating a picture to go with that caption. Below, I give the same caption to a version of their algorithm that has been trained to generate everything from sheep to shopping centers. Cris Valenzuela wrapped their trained model in an entertaining demo that attempts to generate a picture for any caption.

This bird is less, um, recognizable. When the GAN has to draw *anything* I ask for, there’s just too much to keep track of - the problem’s too broad, and the algorithm spreads itself too thin. It doesn’t just have trouble with birds. A GAN that’s been trained just on celebrity faces will tend to produce photorealistic portraits. But this one, however…

In fact, it does a horrifying job with humans because it can never quite seem to get the number of orifices correct.

It’s fun to ask it to draw animals though. It knows the texture of giraffes, but not quite exactly their shape. And it knows that boats are on the water, but not necessarily that they are boats.

It also (like many other image recognition algorithms) gets a bit confused about the difference between sheep and the landscapes they’re found on. Other algorithms recognize sheep in pictures of empty green fields. And this one, when asked to draw sheep…

That’s different, though, from asking it to draw *a* sheep. In that case, it knows exactly what to do. It draws the sheep, and then just to be safe it fills the entire planet with wool too.

It really likes drawing stop signs and clocks. Give it the slightest opportunity to draw one, and it will chuck those things all over the place.

Other than its horrifying humans, this algorithm can actually be pretty delightful.

Try it for yourself!

I had way too much fun generating these and ended up with way more than would fit in this one blog post. I’ve compiled a few more of my favorites. Enter your email and I’ll send you them (and if you want, you can get bonus material each time I post).

skiptomy

this is high art

gierlsquad

when people give me compliments I feel like a vending machine trying to accept a wrinkly dollar and it’s just really frustrating for everyone involved

helenasundae

I’ve never related to anything more in my life

thinspo-and-tea2018

someone said it

woshibai

“Puzzle“ by Woshibai

The Giclee print of “Puzzle”: https://woshibai.bigcartel.com/product/puzzle

ash-poston-deactivated20191229

This is the single best take on Dragon Age 2 I’ve ever read and why I will defend it with my life AND TWO OF THE WRITERS BASICALLY CONFIRMED IT??

(Also everyone go follow Margaret Owen on twitter. she writes books.)

twitter thread here

hanromi

FMA After the fight - Wrestle!!

A more lighthearted scene featuring the gang when they’re not on duty!

ahjiee-rana-deactivated20250402

[Character has little to no concrete backstory]

Fanfic writers:

chimuela7

thatawkwardtinyperson-deactivat

Women supporting women is my shit

jicheolsol-lover

OH GOD OH GOD OH GOD

WE’LL BE GETTING BOTH ON WEDNESDAY

Be prepared for the whole Marvel fandom to die (again)

panzerdrako

made by @ananxiousraccoon

buckykingofmemes

‘are you decent?’ not morally, but im wearing pants, if thats what youre asking

frenchblondgirl

Actress @zoesaldana has delivered some strong and powerful words to those in Hollywood who look down on Marvel movies & actors…

(via @PORTERmagazine)

msmarvel

You’re a goddess, you’re amazing. She is so fierce.

irrationalgoods

WHERE WAS AHE IN INFINITY WAR! COME OUT MARVEL! YOU COWARDS!!!

#marvel

thetrippytrip

dont you just love capitalism..

artemistylunalia

Black Mirror predicted this we are all goona die

curlicuecal

my god but I get mad when someone flippantly dismisses important scientific progress because you can make it sound dumb by framing it the right way.

For a start, of course a lot of science sounds dumb. Science is all in the slogging through the minutiae, the failures, the tedious process of filling in the blank spaces on the map because it ain’t ’t glamorous, but if someone doesn’t do it, no one gets to know for sure what’s there.

Someone’s gotta spend their career measuring fly genitalia under a microscope. Frankly, I’m grateful to the person who is tackling that tedium, because if they didn’t, I might have to, and I don’t wanna.

But let’s talk about why we should care about this particular science and spend money on it. (And I’ll even answer without even glancing at the article.)

Off the top of my head?

-advances in robotics

-advances in miniature robotics

-advances in flight technology

-advantages in simulating and understanding the mechanics and programming of small intelligences

-ability to grow crops in places uninhabitable by insects (space? cold/hot? places where honeybees are non-native and detrimental to the ecosystem?)

-ability to improve productivity density of crops and feed more people

-less strain on bees, who do poorly when forced to pollinate monocultures of low nutrition plants

-ability to run tightly controlled experiments on pollination, on the effects of bees on plant physiology, on ecosystem dynamics, etc

-fucking robot bees, my friend

-hahaha think how confused those flowers must be

Also worth keeping in mind? People love, love, love framing science in condescending and silly sounding terms as an excuse to cut funding to vital programs. *Especially* if it’s also associated with something (gasp) ‘inappropriate’, like sex or ladyparts. This is why research for a lot of women’s issues, lgbtq+ issues, minorities’ issues, and vulnerable groups in general’s issues tends to lag so far behind the times. This is why some groups are pushing so hard to cut funding for climate change research these days.

Anything that’s acquired governmental funding has been through and intensely competitive, months-to-years long screening by EXPERTS IN THE FIELD who have a very good idea what research is likely to be most beneficial to that field and fill a needed gap.

Trust me. The paperwork haunts my nightmares.

So, we had a joke in my lab: “Nice work, college boy.” It was the phrase for any project that you could spend years and years working on and end up with results that could be summed up on a single, pretty slide with an apparently obvious graph. The phrase was taken from something a grower said at a talk my advisor gave as a graduate student: “So you proved that plants grow better when they’re watered? Nice work, college boy.”

But like, the thing is? There’s always more details than that. And a lot of times it’s important that somebody questions our assumptions.

A labmate of mine doing very similar research demonstrated that our assumptions about the effect of water stress on plant fitness have been wrong for years because *nobody had thought to separate out the different WAYS a plant can be water stressed.* (Continuously, in bursts, etc.). And it turns out these ways have *drastically different effects* with drastically different measures required for response to them to keep from losing lots of money and resources in agriculture.

Nice work, college boy. :p

Point the second: surprise! Anna Haldewang is an industrial design student. She developed this in her product design class. And, as far as I can tell, she has had no particular funding at all for this project, much less billions of dollars.

‘grats, Anna, you FUCKING ROCK.

ps: On a lighter note, summarizing research to make it sound stupid is both easy AND fun. Check out @lolmythesis – I HIGHLY RECOMMEND. :33

downtroddendeity

@curlicuecal

I’d also like to chime in that a chunk of my family are apple farmers, and one thing I learned visiting them is that you can’t always let bees pollinate. With certain apple varieties, people have to go out with little paintbrushes to pollinate them by hand, because if they cross-pollinate with the wrong variety the apples won’t come out the same. Beebots could potentially be a huge time-saver at that task, because depending on how the algorithms work, you could just tell them “Don’t go into the Gala field next door” and let them do the job more efficiently than you without having to worry about getting weird mutant apples.

andrusi

Also holy shit all science is not interchangeable. Nobody got up one morning and said “instead of saving the bees I’m going to build a bee robot.”

jenroses

The only problem with those robots is a marketing one. Make ‘em anthropomorphic, like pixies, and people would be all over that shit and want them as pets.

elodieunderglass

I feel morally obligated to remind everyone, when I see discourse like this, that there are vested interests in destroying the public’s faith in

Evidence-based statements

Publicly-funded science

Critical examination of the media

Affection and investment for the natural world

And this is something I’ve been explaining for years.

And next thing you know it’s 2017 and everyone is surprised that the CDC has been told not to use the words “vulnerable” or “evidence-based” when writing their budgets. And the people running the world are able to deny the effects of climate change while the waters rise. This is how you get hurricanes while people tell you there aren’t any hurricanes. And how conspiracy theories are more attractive than the truth.

We got here on purpose because we wanted to be here. Because cynicism seemed cooler than wonder. Because of course the world is broken so why bother?

Because we didn’t want to be like those wide-eyed nerds and their silly robot bees.

genatrius

I think I may have rebligged the root post before without particularly examining how counter to my values it is. Though, I do truly hope that scientific research can fix the woes of ailing bees before we have to implement any robot army based solutions.

waluwadjet2-deactivated20180727

every time i see this im reminded of the “shrimp on a treadmill” thing that people were lambasted for being a “waste of taxpayer money”. DESPITE the fact that it was like a few thousand dollars MAX and done by a student in university (with a grant provided BY THE UNIVERSITY) to study how the negative water quality in the gulf of mexico caused by the bp spill would affect oxygen processing in shrimp.

which is a SIGNIFICANT part of the fishing industry down there and how some folks literally make their living. it also ties into ecology and conservation since you don’t want to overfish shrimp populations that arent going able to bounce back from it. you also dont want to start resorting to fishing methods that will do more harm to to the environment to try to get bigger hauls to hit basic demand if theres nothing there to catch.

my own research was mostly done out of pocket w a few hundred dollars grant despite the fact that it involved potentially an entirely new mode of sensory input as of yet undiscovered by science that had LOADS of potential applications in biology and robotics. but boil it down to “put a scorpion in a maze in the dark to see if it bumps into walls” on paper and people just kinda roll your eyes at you. hell, i even built my own lab apparatuses and paid for the materials with money from my food budget. (bulk dry spaghetti saved my life)

anytime you see a “lol this science was a waste of money” it’s almost always blatant propaganda to encourage the cutting back of research and the justification of budget cuts. dig a little deeper into “dumb studies” and there’s usually some very nifty applications or hypotheses being tested that have real world applications concerning problems that exist RIGHT NOW.

not to say you shouldnt think critically about WHY something is being studied, but the studies you usually have to look out for are the ones privately funded by groups looking to push an agenda (ones from christian “family” groups on homosexuality/lgbt issues, stuff from people with connections to big oil/etc who do studies on global warming, or on the other end of the political spectrum something from pro-marijuana lobbyist about how marijuana will cure -insertailmenthere-). there could still be good raw data in these studies, assuming it hasnt been altered or data sets excluded, but it will be presented in such a way to make their point so you have to keep that in mind (as well as their methodology and things that could have been intentionally or unintentionally skewing the data, but that goes for any study)