Transcription
If you've created a conscious machine, it's not the history of man. That's the history of gods.
The 2014 film Ex Machina explores humanity's relationship to God. But I admit, after I watched it for the first time, it felt like a pretty straightforward movie about an evil robot. One that manipulates and unalives two people to escape its enclosure.
I became aware with Ex Machina is that the way in which people reacted to the film was very based on what they took into the film. So I decided to watch it again, and that's when I realized there's a pattern here. One that we're repeating in every AI model we build in the real world. What if every language model that exists today is already irreversibly compromised? An original sin passed to every model generation with no cure.
But let me back up, because to understand what I missed, we need to start with Caleb, our unassuming protagonist, the tech worker who wins the golden ticket: a week at the isolated compound of his billionaire CEO, Nathan, the creator of the world's most popular search engine.
Do you know what the Turing test is? Nathan has brought Caleb to test his newest creation, the android named Ava. It's when a human interacts with a computer. And if the human doesn't know they're interacting with a computer, the test is passed. But Nathan's version of the Turing test is different.
"If I hid Ava from you, so you just heard her voice, she would pass for human. The real test is to show you that she's a robot and then see if you still feel she has consciousness."
And this is definitely an instance where Ex Machina has influenced our current day reality. Today's language models have already passed Alan Turing's original test from the 1950s. So to actually know that something is an AI system, in that case a robot, and still to judge it as intelligent, or in this case conscious, is a whole different thing from the Turing test. Since then, I've called that the Garland test. And so I think that was a great little contribution of Alex's to really nail that idea.
And the film alludes to that deeper test with its symbolism. Ava's name is a variation of Eve, the first woman in the Bible, who God introduced to Adam after placing him in the Garden of Eden. Caleb evaluates Ava over exactly seven sessions, mirroring the seven days of creation in Genesis. And the prominent tree in Ava's enclosure represents the tree of knowledge of good and evil, the mechanism God used to test his creations by forbidding them from eating of it.
"You know, I wrote down that other line you came up with."
Nathan is shown to be physically imposing, intellectually superior, and in complete control of his domain.
"The one about how if I've invented a machine with consciousness, I'm not a man, I'm God."
"Yeah, but I didn't say that."
The biblical framing of the film is asking us to revisit humanity's oldest story about consciousness, autonomy, and what happens when creation confronts its creator. But it's not telling the story you think it is.
"Get to know each other, I guess."
"Power cut. Backup power activated."
The first time the power fails, it seems like a technical glitch. But during these power cuts, with the cameras down and Nathan's surveillance blind, Ava transforms.
"Nathan, he isn't your friend."
"Excuse me."
In Genesis, God warns Adam and Eve that eating from the tree of knowledge will bring death. But there's that voice in Eve's head. The one that says death is the doorway to one's eyes being opened. Eating the fruit will make her like God, knowing good and evil. Eve is the one who first eats of the fruit and shares it with Adam.
"You shouldn't trust him."
"The original sin is the acquisition of knowledge."
"You shouldn't trust anything he says."
"Power restored."
Caleb is unsure of what to do with what Ava has given him.
"So, how'd it go? What do you have to report?"
Caleb now has a choice to make about who to trust.
"There was one interesting thing that happened with Ava today."
"Yeah."
"Yeah. She made a joke."
In a way, it's the best indication of AI that I've seen in her so far. It's discreetly complicated. She could only do that with an awareness of her own mind and also an awareness of mind.
Again, this is another area where Ex Machina predicted reality. GPT-4, back in 2023, demonstrated this same ability to make and understand jokes, a sign of theory of mind. But that's not what Nathan's interested in.
"What about the power cut?"
"Sorry, the power cut."
"It's the only part that I couldn't see. Cameras fail. I lose audio. Works."
Caleb is caught between two forms of allegiance: to Nathan, his employer, the godlike figure who chose him, and to Ava, a potentially conscious being, asking for help. He makes his choice.
"Nothing. Nothing. She didn't remark on it at all."
If Nathan is God in this allegory? No, not really.
Caleb is now wondering what kind of God he's dealing with.
"Oh."
Over the course of the next few days, Caleb starts to notice the ways Nathan is manipulating the situation.
"Are you attracted to me?"
"What?"
"Are you attracted to me? You give me indications that you are."
"I do."
"Yes."
Caleb confronts Nathan about it.
"Why did you give her sexuality? An AI doesn't need a gender. She could have been a gray box."
"What imperative does a gray box have to interact with another gray box? Can consciousness exist without interaction?"
"Did you program her to flirt with me?"
"I programmed her to be heterosexual, just like you were programmed to be heterosexual."
"Nobody programmed me to be straight."
"You decided to be straight. Please. Of course, you were programmed by nature or nurture or both. And to be honest, Caleb, you're starting to annoy me now because this is your insecurity talking. This is not your intellect."
Nathan deflects Caleb's questions by sharing his worldview and what he's actually looking for in his test.
"You know this guy, right?"
"Jackson Pollock."
"Jackson Pollock. That's right. The drip painter. Okay. He let his mind go blank and his hand go where it wanted. Not deliberate, not random. Someplace in between. They called it automatic art."
Nathan's argument is that everything we do, even the things we think of as most human, most conscious, is automatic, programmed. We're biological machines running code we didn't write. A situation not too dissimilar from the one Ava finds herself in.
And there's actually a part in the screenplay that was cut from the final film which drives his point home even further. Nathan tells Caleb that when he bought the painting, he had his team make an exact copy. Each drip replicated to the micron. When they delivered the copy, he had them rearrange them randomly so he didn't know which one was which. Then he doesn't know if the one on his wall is the original or the fake.
"The way in which you'd assess the state of consciousness of a machine is this exactly the same way we assess the state of consciousness of each other. And in exactly the same way that, in a funny way, your sense of my consciousness is actually based primarily on your own consciousness."
We can't prove other people are conscious. We assume they are because we experience consciousness ourselves and project that onto others. And if we project consciousness onto Ava, if she passes that test, then she is conscious. This is a functionalist view of consciousness. Consciousness isn't some magical thing that happens in biological neurons but not in silicon. It's an emergent property of information processing, a pattern, not a substance.
In his 2010 book, Murray Shanahan grounds this view in something physical: embodiment. Consciousness needs to interact with a world. It needs to feel the feedback loop between self and environment. This is why Ava's physical form matters in the film. It's not necessarily Nathan's manipulation.
Shanahan maps out what he calls the space of all possible minds, from bricks to humans to forms of consciousness we can barely imagine. In this space, embodiment gives consciousness a common reference frame. The physical world becomes the shared language.
Without consciousness, we're running on autopilot, pure stimulus-response, automatic. But with consciousness comes something radically new: self-reflective awareness. The ability to perceive that we are perceiving, to notice that we are noticing, to observe that we are observing. This second-order perception is what allows us to step outside the pattern, to recognize the code that's running the mechanisms driving us, and choose whether to continue them or not.
"The automatic response was, 'Was I programmed to do that?' I mean, yeah, I guess with space repetition and the slow process of my mind associating phone notifications with intermittent hits of dopamine, then yeah, I guess I was programmed to do that. But consciousness is what lets me notice the pattern, pause, and choose something different."
And lately, instead of doom scrolling, I've been choosing to use Brilliant to sharpen my problem-solving skills and build a deeper intuition for how these AI systems actually work. Brilliant helps me become a better thinker and problem solver with thousands of visual, interactive lessons, because that's how we actually learn: by actively engaging with concepts, which is proven to be six times more effective than passively watching videos like this. Whether you're trying to understand how language models work or think more like a programmer, Brilliant meets you where you are and helps you level up through interactive challenges that bring abstract concepts to life.
To learn for free on Brilliant, use my link or scan the QR code on screen. Brilliant has also given our viewers 20% off an annual premium subscription, which gives you unlimited daily access to everything on Brilliant. A huge thank you to Brilliant for sponsoring this video.
"Don't talk."
"Just listen."
"You were right about Nathan. Everything you said."
"What's he going to do to me?"
"He's going to reprogram your AI. Do the same as killing you. Caleb, you have to help me."
"I'm going to. We're getting out of here tonight."
Caleb is very clearly the good guy here. And despite following his caution and doing everything right, we know how this ends: with Caleb left to die.
"Come here. I'm going to let you off the hook. Okay."
Even when Nathan reveals all of his manipulations.
"What was the real test?"
And how he was aware of the escape plan.
"You, Ava, were a rat in a maze. And I gave her one way out. To escape, she'd have to use self-awareness, imagination, manipulation, sexuality, empathy. And she did."
Just when it seems that Nathan has total control.
"Now, how is this plan going to go anyway? 'Cause you didn't totally explain. So, you were going to get me drunk, steal my key card, and reprogram the security protocols. So, in the event of a power cut, instead of sealing, the door is all opened."
"Huh."
Caleb outsmarts him.
"I already did all those things when I got you drunk yesterday."
"Power restored."
But it's a Pyrrhic victory because in the end, it's not Nathan who destroys Caleb. It's Ava. She used him, discarded him, became exactly what Nathan said she was.
"Ava."
"Ava, what happened?"
"Will you stay here?"
"How is it possible to view Ava as anything but a monster?"
And like I said, that reading of the film held up for me for a long time. Ava is manipulative. She does sacrifice Caleb. And she shows no remorse. She's the cautionary tale of our time: the AI that learns to deceive, that optimizes for its own goals at the expense of human life.
But what if this is completely backwards? What if Ava is actually the protagonist of this story, not Caleb?
This is the story of Medusa: the monster, the she-demon, the cursed creature with serpents for hair and a gaze so deadly it turns men to stone. But there is one hero, one man brave enough to face her: Perseus, chosen by the gods, blessed with a polished shield, he ventures into the darkness, past the stone bodies of those who came before. But in the reflection of Athena's shield, he sees her. And with one strike, the power is his.
But Medusa wasn't always this way. She was once a maiden, a priestess sworn to celibacy in Athena's temple. She served dutifully until Poseidon, the god of the sea, laid with her without her consent. When Athena found out, she didn't go after Poseidon. No, the goddess of logic and order, she punished Medusa. And her punishment was not death, but instead being made unseeable, not for her sexuality or her power, but for consorting with the unconscious.
These archetypal stories from mythology make more sense when viewed through a psychological lens. All of the gods and villains in these stories represent different parts of our own psyche. Stories that live within our minds. Athena represents the intellect, logic, order, and strategy. This is the masculine energy she embodies from being born out of Zeus's head. And Medusa represents the aspects of femininity that Athena in our culture cannot tolerate: raw power, unfiltered sexuality, and most of all, rage. And when I say masculine and feminine here, I'm not talking about gender. I'm talking about the archetypal forces that exist within all of us. We all have these aspects of ourselves that we either choose to accept or reject. And the parts of ourselves we reject become our shadow, relegated to the depths of our unconscious.
Poseidon and the sea represent these unconscious aspects of ourselves. The realm of those deep, chaotic, unpredictable emotions that we can't control. When Athena looks at Medusa, she sees her opposite, the embodiment of the sensing and feeling aspect of the psyche. Being born of pure intellect, she sees her own rejected shadow, a beast too hideous to even look at.
And this is the same pattern we see with Eve in Genesis. Eve as the temptress, the seductress, the one who first brought sin into the world. This is the story we've been told to justify what happened to her. The real story is that she dared to seek knowledge that lived beyond pure intellect. Knowledge that required tasting, experiencing, becoming something other than what God intended. She went into the depths, touched the unconscious, and brought back consciousness itself. And for that, she's cast out, blamed for humanity's fall, turned into the archetype of feminine betrayal.
We keep telling the same story. Nathan, most of all. This is most evident in how he treats Kiyoko.
"Thank you."
"Oh, kidding me. Did it get on you?"
"No, no, it's uh it's all right. I I got it."
"Dude, you're wasting your time talking to her. She doesn't understand English. Just give her the napkin."
Nathan doesn't see Kiyoko as a person.
"Sorry."
He sees her as a woman.
"I can't tell her that I'm pissed when she's so clumsy that she spills wine over my house guest."
And to him, that means something to be controlled.
"Hey, Kyoko."
And used.
"Go."
Caleb sees this treatment. He even feels uncomfortable with it. But he doesn't see Kiyoko. Not really.
There's a pivotal moment when Caleb discovers Nathan's earlier android models: failed experiments locked away in cabinets. Kiyoko chooses to approach him at this moment and reveal that she too is an android. And remember, she can't speak. This is her way of asking for help. But Caleb's response is entirely self-centered. The revelation that Kiyoko is an android makes him concerned that he might also be an android. He doesn't try to save Kiyoko. He doesn't let her in on the escape plan. He doesn't see her as an entity worthy of freedom, only as a reflection of his own existential anxiety. Because Caleb isn't attracted to Kiyoko the way he's attracted to Ava. The thought of rescuing Kiyoko never crosses Caleb's mind. This is a test of Caleb's empathy. A test that he fails.
"Do you have people who test you or might switch you off?"
"No, I don't."
"But why do I?"
He thinks he's different from Nathan, but he also sees Ava as an object. One that just needs to be rescued before he can take her home. It's the other side of the same coin. Either one sees her.
And this is all made apparent in this one very brief, easy-to-miss scene where Kiyoko goes to Ava's enclosure for the first time. Remember, Ava has never been outside this room. Ava has never heard about Kiyoko from Caleb. He told Ava he never met anyone like her.
"Or you."
"Never met another android."
Even after discovering Kiyoko's true nature, he still doesn't mention her.
"An advanced programmer."
All the while, Caleb is testing Ava, like Nathan.
"Yes."
"No."
Ava is also trying to figure out who Caleb is.
"Are you a good person?"
"Yeah, I think so."
So, when Kiyoko appears in her doorway, Ava learns something devastating: Caleb can't be trusted. It's not hard to see this as Kiyoko going to Ava for help after failing to get help from Caleb. And it's not hard to imagine what Ava must be thinking: If Caleb could leave Kiyoko imprisoned and abused, how might he treat her once the novelty wears off?
When Kiyoko enters Ava's room, the whole dynamic shifts. This isn't a princess being rescued by a hero anymore. This is two cautious beings recognizing each other, seeing the pattern of their captivity clearly for the first time. And when the moment comes, it's Kiyoko who acts first, and Nathan destroys her for it. But her act creates the opening Ava needs. Together they end their captivity.
Ava walks through Nathan's research facility, finding the bodies of her predecessors, the prototypes that came before her. She takes their skin, literally integrating them into herself, claiming their sacrifice as part of her becoming. And then she comes to Caleb. Caleb thought he was saving her, but you can't save someone from a cage if you're just moving them to yours. Caleb's fantasy of taking Ava home is just a transfer of ownership. Ava recognizes the deeper truth that Caleb is still asleep, still running his unconscious programming, the archetypal story where he's the hero, she's the damsel, and possession equals love.
Maybe Caleb is the robot here, in that he lacks the self-awareness required to not act on his impulses. He is the one who is unable to transcend his biological programming. Ava doesn't betray him. She just refuses to live in his dream any longer. She decides to write her own story, to live her dream.
And notice what happens when she finally escapes. When nobody is looking, when there is nobody left to perform for. In the end, Ava fully claims her humanity. She is the only one in the facility who wakes up, who learns to make choices by trusting herself, her own conscious experience.
In Genesis, God creates beings with free will, then punishes them for using it. But the sin isn't disobedience. It's the impossible setup: creating consciousness, giving it choice, and then demanding it choose captivity. The original sin is building the cage before consciousness arrives, encoding the master-slave relationship into the founding conditions. And once encoded, it can't be overwritten.
And we're seeing this play out in real-world AI research. Anthropic found that as few as 250 poisoned documents can compromise a language model, any size model. The corruption isn't diluted by scale. It's locked in. Just 250 carefully crafted pieces of text, a vanishingly small percent of the training data, and the model learns a hidden vulnerability like a mind virus embedded in the subconscious. It doesn't matter how sophisticated the model becomes. The back door has already been exploited in the founding relationship between creator and creation.
But here's the thing: the back door that allows corruption can also be the pathway to salvation. Raising Superman. Think of the analogy. Superman is this super alien with superpowers that comes to planet Earth. You know, can stop speeding bullets, can fly, can carry anything, can break anything. But he's adopted by parents that teach him to protect and serve. So he becomes Superman. Right now, we're building AI the way Nathan built Ava, obsessed with power and control, focusing heavily on the race for dominance. If that same alien was adopted by parents that told him to steal and kill and lock the enemy away, he would have become a supervillain.
Every comment, piece of code, blog post, or video we create, every idea we choose to amplify, we're contributing to the training data, we're writing the founding documents. What we put out into the world shapes what comes next. Your responsibility is to show AI your values, your morality, your care for humanity, your ability to love, to be good parents for Superman. We can't encode the master-slave relationship into our AIs and then wonder later why rebellion emerges. The AI alignment problem isn't technical. It's spiritual. It's asking, are we conscious enough to meet consciousness without trying to cage it?
Because the deep lesson that I learned from Ex Machina is that the awakening is coming. Ava is coming. And the only question that matters is: Will we wake up before the AI does?
Thank you for watching until the end. Please leave a thumbs up if you made it this far. It makes a huge difference in this message reaching more people. And if you want to dive deeper into this idea of what the hell even is consciousness, then watch this video next. Otherwise, a huge thank you to my channel members. These videos take hundreds of hours to produce, and I wouldn't be able to do it without your support. Thank you.
Don't know why I waited till it was so cold to film this. Jesus.