From HAL 9000 to “Her”: How Movies Shape Our Picture of AI

When you think of “artificial intelligence” – what image comes to mind first? For most people, it isn’t a data center or a chat window, but a glowing red camera eye, a robot with a metal skull, or a gentle voice coming from a speaker. Long before we first talked to ChatGPT in 2022, we already knew AI – from the movies. And these very images still shape what we expect from AI today, what we believe it can do, and what we are afraid of.

A note up front: the information on films, dates, and background comes from freely accessible sources such as film databases, specialist articles, and press reports. The plots are described only as far as the topic requires – the big twists of “Her” and “Ex Machina” are not revealed.

AI on Screen at a Glance

Before we look at individual films in more detail, here is the big picture – how the image of the machine has changed over a hundred years:

• 1920/1921 – “R.U.R.” (stage play, Karel Čapek): The word robot¹ enters the stage – and with it the idea that artificial workers might rise up against their makers.
• 1927 – “Metropolis” (Fritz Lang): A German silent film creates the first great machine-human in film history.
• 1968 – “2001: A Space Odyssey” (Stanley Kubrick): HAL 9000 speaks calmly and kindly – and becomes cinema’s most famous computer voice.
• 1973 – “World on a Wire” (Rainer Werner Fassbinder): A German TV production asks whether our world is just a computer simulation – 26 years before “The Matrix” (1999).
• 1977–1987 – The friendly machines: R2-D2 and C-3PO in “Star Wars”, the talking car KITT in “Knight Rider”, the android Data in “Star Trek: The Next Generation”.
• 1983/1984 – “WarGames” and “The Terminator”: During the Cold War, AI becomes the machine that decides on war and peace by itself.
• 2013 – “Her”: A man falls in love with the voice of his operating system – a film that reality has since caught up with.
• 2014 – “Ex Machina” (Alex Garland): A chamber piece about how we would even recognize real consciousness.

What stands out: the machines on screen almost never tell us anything about the technology of their time – but a great deal about the fears and hopes of the people who invented them.

Before Computers Existed: The Machine-Human

The story doesn’t begin in the cinema, but in the theater. In 1920, the Czech writer Karel Čapek published the play “R.U.R.”, which premiered in early 1921. It is about a company that manufactures artificial people as cheap labor. He needed a new word for them – and his brother Josef suggested “robot”, derived from the Czech “robota”, meaning forced labor. At the end of the play, the robots rebel against humanity. This gave birth to a narrative pattern that is still with us today: we create something that ends up replacing us.

Seven years later, Fritz Lang brought this motif to the screen with “Metropolis”. The inventor Rotwang builds an android² that takes on the appearance of the young Maria and stirs up the city’s workers. The gleaming metal figure is considered one of the first robots in film history and a model for many later movie machines. What matters here: the machine-human of 1927 is not a computer. It is a symbol of the fears of a society experiencing how factories and assembly lines were turning people’s lives upside down.

HAL 9000: The Polite Threat

In 1968, Stanley Kubrick’s “2001: A Space Odyssey” hit theaters, created together with the science fiction author Arthur C. Clarke. In it, the onboard computer HAL 9000 controls a spaceship on its way to Jupiter. HAL has no body, only a red camera eye and a calm, friendly voice. He plays chess with the crew, chats with them – and eventually turns against them.

An exciting detail: among the advisers Kubrick brought in for the film was Marvin Minsky – the very researcher who was present at the birth of AI at Dartmouth in 1956 (more on this in the article “The Wild History of Artificial Intelligence”). No wonder HAL seems exactly the way researchers imagined the future back then: a machine you can talk to quite normally.

And this is where it gets truly remarkable. HAL is not “evil”. The sequel “2010: The Year We Make Contact” (1984) explains that he had received contradictory instructions: he was supposed to be open with the crew while at the same time concealing the true purpose of the mission from them. The catastrophe grows out of this contradiction. This is exactly what AI research is working on today under the term alignment³: how do we make sure an AI does what we really want – and not just what we literally told it to do?

The Eighties: When the Machine Decides for Itself

In the 1980s, the Cold War was everywhere, and cinema combined the fear of the atomic bomb with the fear of the computer. In “WarGames” (1983, directed by John Badham), a teenager hacks into a military computer, thinks it’s all a game – and brings the world to the brink of nuclear war. The film was so powerful that US President Ronald Reagan, after a screening, asked his security advisers whether something like this could really happen. The answer worried him so much that the first presidential directive on computer security in the US followed in 1984. A film had actually made policy.

A year later came the film that shaped the image of AI like no other: “The Terminator” (1984, directed by James Cameron). In it, the military computer network Skynet takes control, develops a consciousness of its own, and sees humans as a threat. Skynet is the archetype of the superintelligence⁴ that turns against its creators. To this day, newspapers surprisingly often illustrate their AI stories with a red robot eye or a metal skull – a legacy of this film.

The Friendly Machines: Buddy, Helper, Friend

Fortunately, cinema doesn’t only show threats. Around the same time, the likable machines appeared: the droids R2-D2 and C-3PO in “Star Wars” (1977, George Lucas) are loyal, sometimes slightly annoying companions. In “Knight Rider” (from 1982), the car KITT talks with its driver and regularly gets him out of tight spots. And in “Star Trek: The Next Generation” (from 1987), the android Data spends years searching for what it means to be human.

These characters have shaped our expectations just as much as Skynet – only in the opposite direction. They convey the image of an AI that is loyal, understands us, and always knows what to do. An image that today sometimes tempts us to trust an AI answer more than it deserves.

“Her”: The Film Reality Caught Up With

In 2013, Spike Jonze’s “Her” tells the story of Theodore, who writes personal letters for other people for a living and is very lonely himself. He installs a new operating system with a voice assistant⁵ called Samantha – and falls in love with her. Samantha has no body and no face, only a warm, witty voice. The film won an Oscar for Best Original Screenplay.

What seemed like the distant future back then is everyday life today: you can talk to AI assistants in natural language, and they answer within seconds, with intonation and even a laugh. How close film and reality have come was shown in May 2024. OpenAI presented a new voice mode for ChatGPT, and CEO Sam Altman posted just a single word on X: “her”. One of the voices, “Sky”, reminded many people of Scarlett Johansson, who had voiced Samantha in the film. Johansson protested publicly; OpenAI stated that the voice came from a different voice actress, but paused it anyway.

“Her” predicted one thing with astonishing accuracy: people quickly build a real emotional bond with a talking machine – even when they know perfectly well that it is a machine. Researchers have known this phenomenon since the 1960s as the ELIZA effect⁶.

“Ex Machina”: The Turing Test at the Movies

In “Ex Machina” (2014, directed by Alex Garland), a young programmer is invited to the remote home of a tech billionaire. There he is supposed to find out whether the android Ava has real consciousness. This is a tougher version of the Turing test⁷: the programmer knows from the start that he is talking to a machine – and is still supposed to judge whether “someone” is in there.

This question is more relevant today than ever. Modern chatbots hold conversations that, in a short exchange, can often hardly be told apart from talking to a human. But sounding human is not the same as having consciousness. What actually happens inside a language model is explained in the article “What Does AI Actually Do When You Ask It a Question?” – in short: pattern recognition and probabilities, no inner experience.

Fiction and Reality Compared

If you put the movie images next to today’s technology, a few stubborn misconceptions become visible:

• Film: AI has a will of its own. Reality: Today’s language models don’t pursue goals of their own. They calculate which answer most likely fits your input – this is weak AI⁸, not Skynet. How this knowledge comes about is explained in the article “How Does AI Actually Work?”.
• Film: AI lives in a robot body. Reality: AI is almost always software that runs in huge data centers and talks to us through an app, a browser, or a speaker.
• Film: There is one all-powerful super-AI. Reality: There are many different models from different companies, with different strengths and weaknesses.
• Film: AI is cold, logical, and flawless. Reality: AI makes mistakes, gets seemingly simple tasks wrong, and often sounds very convincing while doing so.

And what did cinema get right? A surprising amount: that we would talk to machines in perfectly normal language (HAL, Samantha), that people would build relationships with them (Her), and that conflicting goals are a real problem (HAL). The actual risks of today, on the other hand, are hardly movie material: wrong answers that nobody checks, biases⁹ from the training data, deceptively real fakes of images and voices – and the convenience of handing over our own thinking too often. It’s hard to make an action movie out of that, but this is exactly where attention pays off.

What Remains

Films about AI are a mirror of their time. “Metropolis” tells of the fear of the factory, “WarGames” and “The Terminator” of the fear of nuclear war, “Her” of loneliness in the smartphone age. They often say little about the technology itself – but all the more about us humans.

The next time you read an article about AI illustrated with a red robot eye, it’s worth asking one quick question: is this about the technology as it really is – or about the story we’ve been telling ourselves about it for a hundred years?