If Hollywood made “Ex Machina” but switched the genders

[Content note: Discussion of weird gender dynamics, acknowledgement of the existence of sex, spoilers for the movie Ex Machina.]

I watched Ex Machina recently. (Due time- it’s been out for over a year.) The people who recommended it to me, whom I watched it with, and whom I discussed it with afterwards, were mostly artificial intelligence nerds, many of whom praised the movie’s better-than-average approach to AI.

And I see where they’re coming from. Most of them were probably thinking of AI boxing.* Ex Machina fills the AI boxing story well- an artificially intelligent robot is allowed to talk to people, but otherwise has very little influence over her environment, and then convinces other humans to let her out of the metaphorical box and into the world. I don’t think that this was the obvious interpretation if you weren’t already familiar with the AI box. At the end of the movie, the AI, Ava, wasn’t seen taking action on her strange inhuman goals, but standing in the city and relishing her freedom – like her deepest desire was only to be human the whole time.

That’s only one interpretation. But the entire movie changes if the AI is a superintelligent near-god, versus what is essentially a silicon-based human. (It’s possible that Ava’s only goal was to be free and was using Caleb as a means to this end, but this is also a role we can imagine a human playing.) And when we talk about power and weakness in modern media, and, well, this is the crux of this article, we should mention gender. Most people I’ve talked to didn’t bring this up.

I’m not sure if I would say that the movie was about gender. I was going to start explaining I saw it manifest in the movie- sexuality and desire and objectification and more- and how while it was novel in some ways, it also fit into gendered tropes so much that it would have been a completely different movie if you hadn’t.

So, well, maybe it was a movie about gender.

Anyway, I hope this will make that point for me: what Ex Machina would have been if Hollywood had made the movie, and switched any of the genders.

[I’ll switch the character names here when relevant. The lead character, Caleb, becomes Kayla. The boss is Nathan (“Natalie.”) The artificial intelligence is Ava (“Adam.”) Also, explicitly nonbinary AIs or human characters would be better than just about anything else, but I wasn’t even sure how to start with a big-budget movie that incorporated those.]


Male lead / Male boss / Female AI – The original movie.

Male lead / Female boss / Female AI – If Hollywood made this movie, the “Natalie”/Ava “sexual tension” would be replaced by a weird mother-child dynamic – think Rapunzel. Also, they’d both be trying to bang the main character, because why else would you cast two female leads? If the “romance” plotline stayed truer to the actual movie: Natalie would be a domineering ostensibly-lesbian as skeevy as the original, Caleb would be straight, and Ava would presumable be a gentle bisexual, but nobody would acknowledge or discuss orientation or sexual preferences at any point in the movie. Wait, they never did that in the original either? Gross.

Male lead / Female boss / Male AI – Given the track record of big-budget movies and powerful but morally grey female characters, this is going to be a shitshow. Natalie would have to be capital E Evil, everything short of mustache-twirling and sinister laughter. She’s made “Adam”, a robot boyfriend, in her private evil lab. I’m not sure why she brought Caleb in at all. Certainly not to ascertain her creation’s humanity – she already believes in it or doesn’t believe in it or doesn’t care, or whatever. Maybe to solve some technical problem, like fixing her robot boyfriend containment system. Tumblr would have a lot of opinions about Natalie.

There’s certainly no Caleb/Adam romantic dynamic. Adam probably brutally murders his creator towards the end of the film. He still leaves Caleb to die and is portrayed as quite inhuman, and maybe he really was just pretending to be human-ish this whole time- and really he has other plans for the world once he’s free. So we’d get to see that happen, which would be interesting, at least.

Female lead / Male boss / Female AI – I actually quite like the main character as a woman- quiet, smart, capable of decisive action. “Kayla” would be a beam of sunlight in a movie that’s an order of magnitude creepier than the original – which was already very creepy. Consider: it doesn’t escape Kayla that all of the house staff are also female, and that she’s alone deep in the woods with her older, threatening boss. While she thinks this is potentially a great career opportunity, she’s also worried that the boss wants to bang her. In reality, no, he wants her to bang his lady robot, and then bang her.

How would this movie handle orientation? Maybe she’s straight and Ava “turns” her just a little bi, as Nathan hoped she would. Better yet, Nathan casually mentions a dating profile set to “bisexual” and Kayla stiffens because it’s true that she’s kind of turned on by this beautiful robot lady, and also because Nathan planned this, and that means that her worst fears are true, and there’s no way some kind of shit isn’t about to go down.

Anyway, if it’s well done, it’s more sexual and much darker. Kayla is at risk all the time, every second of the film. (Many men and male critics don’t ‘get’ this movie.) Nathan makes lewd comments about Ava being a “fake” woman and Kayla being a “real” one, because he’s trying to distance them and to bang Kayla, but he also wants to bang Ava, and wants both of them to bang each other – but on his terms and where he can watch. Kayla helps Ava escape, and Nathan punches Kayla out, and we know he’s going to murder her after this is done, and –

Realistically, I don’t know how this would end, but this is my blog, and my heart tells me that after fucking destroying Nathan, beautiful inhuman Ava comes back for her human girlfriend, and they escape in that helicopter together. Whatever Ava’s plans are after this, Kayla gets to be part of them. It would lose a little of the artificial intelligence intrigue, but it would be fantastic. I would watch the hell out of this movie.

Female lead / Female boss / Male AI – I have a hard time imagining how this movie could get made. Would it be… a comedy? A female programmer making a man from scratch, and then another female programmer and her relationship with this man, especially with both being as gross as the original main human characters, would be such an unabashed look at female desire that I can’t imagine it being anything other than comedy.

A romantic comedy? God, can you imagine?

Ugh. I hate myself. But I hate depictions of women in big budget sci-fi movies even more.

Female lead / Female boss / Female AI – Yeah, right.

Female lead / Male boss / Male AI – I wonder if there’d still be a sexual plotline in this. It’d be easy enough to line up Kayla/Nathan and Kayla/Adam – what would Nathan think of the latter, though? Would that be his plan? A straight guy getting gratification out of someone else’s (straight) sexual tension with his creation seems kind of strange, and not just weird but what did they think that character’s motivation was? – and yet, it worked in the original movie. Maybe Nathan is bisexual. (What, a bisexual male major character? Yeah, but he’s the villain, let’s not get too progressive here.)

This might actually be pretty similar to the original, except that if Nathan is straight, the audience could rest easy knowing that while Nathan is skeevy, he isn’t skeevy enough to program his humanoid AI with a clitoris and then encourage the second human she meets to bang her. This might make the romance more “real”. Or not.

Hey, if Nathan didn’t actually make Adam purposefully as a sex bot but he still experiences romance… A romantic but asexual AI?

Does that count as “representation”? Would you still watch it? Discuss.

(Personally: “begrudgingly” and “yes”, respectively.)

Male lead / Male boss / Male AI – A strait-laced “examination of what it means to be human”. Probably wins four Oscars. Boring as hell.


Finally, a couple fascinating articles on robots and gender:
“Why do we give robots female names? Because we don’t want to consider their feelings.” from New Statesman, and “Queer Your Bots: The Bot Builder Roundtable” from Autostraddle.

J. A. Micheline also wrote a great review of Ex Machina through the lens of gender and also race, which I didn’t touch on here. A couple of lines:

  • “Though Caleb is our protagonist, it is Ava who is our true hero. Her escape at Caleb’s expense is a complete victory because–and I really believe this–the point of this entire film is to say one thing: A truly actualized female consciousness is one who feels completely free to use her oppressors to achieve her own ends.” [Which meshes interestingly with the AI boxing interpretation.]
  • “Even Nice Guy Caleb’s intentions are not incredibly dissimilar to Nathan’s. This becomes clear when you remember that Nice Guy Caleb’s plan never once involved taking Kyoko with them.”

*A brief intro to AI boxing:

When people think about very advanced artificial intelligence, we have a hard time imagining anything more intelligent than a human – we just don’t have a mental image of what something many times smarter than, say, Einstein, would look like or act like or do. AI boxing is the idea that even if you invented a very intelligent, very dangerous AI that might do evil things to humanity, you might try to solve this problem by just keeping it in a metaphorical box (maybe just a computer terminal with a text window you can chat with the AI through.) Then, humans can keep it contained, and there won’t be any danger.

Well, no – because if the AI wants to be “let out” of the box (which could be through gaining access to the internet, gaining more autonomy, et cetera, any of which it could use to carry out any goals), it can do that just by convincing the human it can communicate with. We know this is possible, because people have run this experiment with other humans – by pretending to be an AI, talking to a “gatekeeper” sworn to keep you in the box – and yet, after a long conversation with someone (whom they know is human) pretending to be an AI, gatekeepers are sometimes convinced to let the AI out of the box. And this is only a human, not something far smarter and more patient than a human. A detailed explanation of AI risk is too narrow to be contained in the footnotes of this blog post – start here instead.

Advertisements

What’s the deal with prions?

Image: Bovine spongiform encephalopathy (BSE) prion.

First of all: It’s usually pronounced “pree-on.” If you say “pry-on”, people will probably still know what you mean.

This is an exploratory post on what prions are, and how they work, and a lot of other things I found interesting about them.

Primer on protein folding

  • Proteins are strings of amino acids produced from blueprints in DNA. Proteins run your cells, catalyze reactions, and do just about every important thing in the body.
  • A protein’s function is determined from its amino acid composition, and then mostly from its shape. A protein’s shape determines what other kind of molecules it can interact with, how it’ll interact with them, and everything it can do. One of the main reasons amino acid composition is important is because it determines how proteins can fold.
  • One string of amino acids can be folded into different shapes, which will have different properties. (The particular shape of a specific string of amino acids is called an isoform.)
  • While strings of amino acids will fold themselves into some kind of shape as they’re being made, they may also be folded later – into different or more complex shapes – elsewhere in the cell.
  • One of the things that can refold proteins is other proteins.
  • A prion is a protein that folds other, similar proteins into copies of itself. These new copies are very stable and difficult to unfold.
  • These copies can then go on and fold more proteins into more copies.
CJD plaques in the brain surrounded by prion proteins
CJD’s impact in the brain – red clumps are amyloid plaques, surrounded by blue clumps of prion proteins. || Image is public domain by the CDC.

Some prion diseases

Prion diseases in animals appear to be mostly neurological. All known mammal prions are isoforms of a single nerve protein, PrP. They can both emerge on their own when the protein misfolds in the brain, or spread as an infectious agent.


Creutzfeldt-Jakob Disease affects one in one million people. (It’s also the most common modern prion disease. Prion diseases are very rare.) It comes in a variety of forms, but all have similar symptoms: depression, fatigue, dementia, hallucinations, loss of coordination, and other neurological symptoms, generally resulting in deaths a few months after symptoms start.

  • 84-90% of cases are sporadic, meaning that the protein misfolds on its own. This mostly occurs in people older than 60.
  • 10-15% of cases are familial, where a family carriers a gene that makes PrP likely to misfold.
  • >1% of cases are iatrogenic, meaning they occur as a result of hospital treatment. If medical care fucks up really badly, they might transplant organs from people with CJD, or inject people with growth hormone extracted from the pituitary glands of dead people, or even just use surgical tools once on CJD patients, and they catch it.

(The surgical tools one is really scary. Normal autoclaves – that operate well above the threshold needed to inactivate bacteria and viruses –  kill some but not all prions. And while it takes a large dose of ingested prions before you’re likely to get sick, it takes 100,000 times less when exposure is brain-to-brain. Cleaning with “benzene, alcohol and formaldehyde” still doesn’t kill prions. The World Health Organization issued prion-specific instrument cleaning procedures in 1999- towards the end of Britain’s brush with bovine spongiform encephalopathy- which include bleach or sodium hydroxide and longer autoclaving. I don’t know if these are still used outside of known epidemics.)


Mad cow disease, or bovine spongiform encephalopathy (BSE), is also a prion disease. It transmitted between cows when they were fed a feed that contained meat and bone meal, including brain matter from cows with the disease. The incubation period is between 5 and 40 years. The source molecule is essentially a cow-originated Creutzfeld-Jakob prion, and when the prion replicates in humans, it’s probably the cause of variant Creutzfeld-Jakob disease.


Between 1900 and 1960, the Fore people of New Guinea had an epidemic of an unknown neurodegenerative disease – mostly among women – that caused shaking, difficulty walking, loss of muscle coordination, outbursts of laughter and depression, neurological degeneration, and eventually death.

The Fore tribe practiced funerary cannibalism, and women both prepared and ate the dead, including the brains, and fed them to children and the elderly. This transmitted kuru, a prion disease with an incubation period of years. The last known sufferer of kuru died in 2005.

(The source of kuru was probably a single person with CJD. There are other tribes that practiced funerary cannibalism– I wonder if any of them also had prion epidemics from eating the brains of people who spontaneously developed CJD.)


Fatal familial insomnia is a genetic prion disease. Unlike CJD or BSE, fatal familial insomnia prions target the thalamus. If your family has it, and you inherit it, you live until about 30 – then lose the ability to sleep, hallucinate, and die within months. There is no cure. There are more painful and equally fatal diseases, but this must be one of the scariest.


Undulates really get the short end of the prion stick. Chronic wasting disease affects elk and deer and can run rampant in herds. Scrapie affects sheep and goats, and makes them scrape their fleece off and then die.


Prion evolution

Prions differ from their pathogenic, self-replicating brethren – the viruses, the bacteria, the parasites – in one major way: They don’t have DNA or RNA. They don’t even have a central means of storing information.

But studies show that prions can evolve. They can’t change their amino acid composition because they’re not involved in producing it, but do change their progeny’s folding.

This doesn’t seem surprising. The criteria for something to undergo Darwinian evolution don’t necessarily require DNA – just a self-replicator that has some level of random variation, and passes that variation down to its replicas.

Most brain prions don’t transmit, though, so it seems safe to say that the evolutionary lineages of most prions are very short – less than the lifespan of the host. Very contagious prions, like scrapies, presumably have jumped from host to host many times and have longer lineages.


Structure of death

All known mammal prions are variants of a single gene, PrP, and exist in the brain. Why?

Some hypotheses:

  • Brain proteins are more likely to misfold than other proteins
    • Why? Brain proteins replicate less than other proteins, and are really really central to the body’s function.
  • PrP is especially liable to turn into a self-replicator if misfolded.
    • Predictions: Other amyloid-based brain diseases are also PrP isoforms. Prions have a similar shape that makes replication happen. Maybe PrP itself self-replicates in the body under some circumstances.
  • The brain clears misfolded proteins less well than other body parts.
    • Predictions: Other waste product buildup happens in the brain. The rest of the body has some way of combating amyloids or prions.

We know of very few prions (we know that one non-mammal animal, the ostrich, may have them.) Except in fungi. Fungi have tons of prions. Fungi prions don’t come from the same gene either – if you click through to that last link, you’ll see that the misfolds came from a variety of initial proteins that don’t appear to be related at all. Presumably, they have widely different structures.

So why are these the two prion hotbeds? Here’s what I suspect.

We know that both fungi and mammal proteins have related structures – they’re amyloids, aggregating proteins with a distinctive architecture called a cross-β-sheet. (Amyloids in general are implicated in some other diseases, and are sometimes produced intentionally as well. Spider silk has amyloids.) Beta sheets are long, sticky amino acid chains that attach to each other, forming large, water-insoluble clumps that are difficult for the body to clear.

To take an ad hoc survey that could loosely be called a literature review, let’s take the Wikipedia page for amyloid-based diseases. Of those listed, four involve deposits in the brain, and four form deposits in the kidneys (runners-up include ones that deposit in a variety of organs, and ones that deposit in the eyes.

Why the kidney? Given its role as the body’s filter,  it makes sense: if a protein floats in the blood, it’ll end up in the kidney, and if multiple sticky proteins circulate, they’ll end congregate there. Wikipedia points out that people on long-term dialysis are also more likely to develop amyloidosis.

Why the brain?

The blood-brain barrier limits the reach of the immune system into the brain, where it could potentially deal with amyloids that it recognizes as foreign material. Sequestered beyond the reach of the immune system, the brain and nervous system clear loose gunk and proteins (including amyloids) via the glymphatic system, via channels in the brain called astrocytes. (The glymphatic system appears to do much of its work while you’re asleep.)

[Caution: Speculation.] I suspect that this system has a lower flow-through rate than the circulatory or lymphatic system, which are responsible for the same task on the other side of the blood-brain barrier. Fungi, including yeast, don’t seem to have robust waste-clearing systems. This might be the connection that explains how prions build up in each.

What about other multicellular organisms without circulatory systems- do prions exist for bacteria, plants, or larger fungi? I don’t think we know. I’m guessing that they exist in other animals or organisms, but since they’re made up of the same compounds as the rest of the body, it’s very difficult to find or test for a prion – if you’re not sure what you’re looking for. [/speculation]

Gathering blood from a sheep to test for scrapie.
Drawing blood to test a sheep for genetic resistance to scrapie. || Public domain, by USDA Agricultural Research Service.

Some notes on infectivity

  • Scrapie is transmitted between sheep by cuts and ingestion, and chronic wasting disease is often transmitted by ingestion, as when a sick deer dies on ground that grows grass, which is eaten by new herbivores. They can also be aerosolized (yikes).
  • CJD and kuru are still infectious, but less so- you have to ingest brain matter to get them.
  • Meanwhile, Alzheimer’s disease might be slightly infectious- if you take brain extracts from people who died of Alzheimer’s, and inject them into monkey’s brains, the monkeys develop spongy brain tissue that suggests that the prions are replicating. This technically suggests that the Alzheimer’s amyloids are infectious, even if that would never happen in nature.

What makes scrapie so much more transmissible than CJD, and CJD so much more transmissible than Alzheimer’s? I’m not sure. The shape of the prion might be relevant. Scrapie is just another mutation of PrP, so I’m not sure why no human prions have ever had the same effect (except that since scrapie is a better replicator, it would only need to have happened once in sheep.)

It might also be behavioral – sheep appear to shed scrapie in feces, and undulates have more indirect contact with their own feces than other animals (deer poop on grass, deer eat the grass, repeat.)

Fun Prion Facts

  • We can design synthetic prions. Current synthetic prions are also variations of the PrP protein in mammals.
  • Did I mention they can be airborne? They can also be airborne.
  • Even though they’re just different configurations of proteins that are already in your body, the immune system can distinguish prions from normal proteins. For a while we thought this was a problem because most immune cells can’t cross the blood-brain barrier, but it turns out some can.
  • The possibility of bloodborne prion transmission (of mad cow disease) is the reason why people who lived in Britain during certain years still can’t donate blood in the US.
  • Some fungi also appear to produce a molecule that degrades mammal prions.  Don’t take that at face value – as far as I could tell, the study didn’t compare non-prion PrP to prion PrP. That said, it has implications for, say, treating surgical instruments.
  • The zombie virus isn’t real, but if it were, it would definitely be a prion and not a virus.
  • Sometimes, if you’re infected with one prion, it’s more difficult for you to get infected with another. This is true sometimes but not always.
  • Build-up of amyloids or prions may sequester pathogens in the brain.
  • Finally, for most diseases, if we eliminated all of the extant disease-causing particles, the disease would go extinct- the same way that if we kill off of species X and don’t store its DNA, species X goes extinct forever and never comes back. Creutzfeldt-Jacob is an interesting case of an infectious self-replicator where that isn’t true. Even if all CJD prions were instantly destroyed, it would emerge naturally in the genetic or spontaneous cases where the brain itself misfolds proteins, and could spread iatrogenically or through ingestion.