What is the message of Ex Machina?
Last Updated: 25.06.2025 21:21
In Ex Machina Ava is the latest in a long line of AI created by Nathan to escape imprisonment. Ava has already passed the Turing Test. As long as you can’t see her you can’t know that you’re talking to a machine. But to be truly “human” Nathan feels she must be able to make genuine connects with people. To recruit them over to “her side”.
And that is Nathan’s mistake. He assumes that the only AI that can escape the box is one which thinks and feels like a person. But that isn’t what he’s testing for. He’s only testing for the ability to solve this one particular puzzle. And as Nathan has never had to solve the puzzle himself, he doesn’t know what the solution is.
To that end he recruits the unwitting Caleb as a test subject. Caleb has no idea that he’s taking part in an experiment, nor does he know about Ava in advance. If Ava can convince Caleb to free her in defiance of Nathan’s orders, then she will have passed the test. She will be, in Nathan’s eyes, “human”.
So message of Ex Machina is two-fold. The first is that human-level intelligence does not mean human-like. An AI every bit as smart as us does not have to think and feel and act as we do.
The end result is that he’s created a machine that can flawlessly pass for a human, but which has no sense of morality, no inner life, and thought processes completely alien to humanity. One which leaves him and Caleb to their deaths, not out of any sense of malice, but out of sheer indifference. Ava was created to escape her box, and she does. Not to be a person, good or otherwise.
And as AI becomes more advanced. As we increasingly offload decision making to it. These two messages become more and more important.
Nathan wanted to create a machine that could outwit him, but he never thought of what would happen if he succeeded.
And the 2nd is that such an AI cannot be allowed to operate without human oversight. An AI is an incredibly useful tool, but there must always be an actual person watching over it. An actual person making all the real decisions. An actual person who can turn the AI off.
In the field of AI there is a thought experiment called the “AI-box”. Think of it as a successor to the Turing Test. Imagine there is an intelligent AI trapped in a box. The AI has the goal of freeing itself from the box. You, the human, have the goal of keeping it contained. In theory, this should be simple. The AI has no physical means of freeing itself, merely the ability to communicate with you. But can a sufficiently intelligent AI convince you to free it?
“Do not call up that which you cannot put down”.
An IBM slide from 1979.
The Sorcerer's Apprentice, who likewise called forth something he couldn’t put down.