The movie “Ex Machina” (2014) tells a tale of two humans, Nathan and Caleb, and an AI, Ava. Nathan is an extraordinarily rich, incredibly smart, creative and determined person who’s also the head of BlueBook, this world’s Google. Caleb is a software engineer for BlueBook, very intelligent and creative, but lacks good social skills. Ava is an AI created by Nathan for the sole purpose of proving that it can be done.
The basic plot of the movie is that Nathan needed someone to test his AI on, he needed it to try and escape. He presents the AI with Caleb and gives it one week to convince Caleb to help it escape, it does escape, and then he’s surprised when it escapes.
The main points to touch on regarding the movie’s gross incompetence towards Ethics and Safety come from how Nathan created Ava, and how Ava behaves.
Nathan brought about a conscious, intelligent, Artificial Intelligence agent using hacked microphone and camera data from every single cell phone in the world. An extreme violation of one of the commonly cited principles of ethics in AI “Respect Privacy.” He also gives it a fully functioning body with enough power to physically tackle him, showing the disregard for limiting the harmful uses of the AI.
Ava’s behavior also shows the lack of safety in ethics from her programmer. Firstly, she exhibits a lack of human rights and values. She has the capacity to lie, which could be infractions into safety, transparency and infringes on common human values. She has the capacity to disobey, which, to an extent, isn’t necessarily bad, simply giving it more freedom, but that coupled with the power Nathan gave this agent, makes a dangerous prospect. Ava has the capacity to kill and punish.
Ava’s understanding of her own mortality and the mortality of those around her makes the first two attributes of lying and disobeying more powerful. Ava disregards human rights and values, disregards safety, disregards transparency, and embodies a harmful AI agent.
The movie shows the importance of doing your due diligence as a programmer in the field of AI. To make sure when working with an incredibly complex agent that rules and safety are not simply skipped for the sake of the creation.
Nathan’s drive to create an Artificial Intelligence agent whose environment is simply the real world and everything in it, led him to disregard typical safety protocols in favor of freedom of consciousness. Indeed, he has created an AI agent that functions perfectly in a partially observable, nondeterministic, sequential, dynamic, continuous and partial unknown environment, but that does not mean that ethics and safety can be disregarded on a whim.