Movies Reveal Our Emotional and Irrational Relationship with Artificial Intelligence

15.05.2024|Christian Kreutz

Movies are a great mirror of the Zeitgeist and illustrate common fears or desires of societies. I recently watched the movie "Civil War," and it is frightening to consider a what-if scenario where the USA slides into a civil war. It paints a gruesome picture of how quickly moral standards can decay.

Can movies also help to raise broader awareness of the challenges posed by technology?

In the 80s, the movie "War Games" illustrated the dangers of cyber attacks. It allegedly influenced the Reagan administration to implement the first cyber regulation in the USA.

This what-if scenario is great for picturing technological advancements that are so difficult to depict, particularly during moments of hype. As the saying goes, we overestimate the short-term effects of new technologies and underestimate the long-term effects. Artificial intelligence is exactly such a topic. Only a few years ago, in the movie "Her," a relationship with a virtual partner was absolute science fiction. Now, on platforms like Replika, people build "friendships" with chatbots.

As much as movies allow us to glimpse a possible future, they can easily give us false illusions about artificial intelligence. This is illustrated by the fascinating results of PhD work by German researcher Nadine Hammele (Video presentation German only), who analyzed 71 movies on artificial intelligence.

In her analysis, Hammele found that almost all AI movies exhibit a trend of extreme humanization of AI. This is probably also a reason why people are getting wowed by large language models, as their replies feel intelligent and empathetic. Hammele discovered that in nearly all movies, AI is portrayed as a strong intelligence equivalent to humans in both analytical and social intelligence.

There is a notable scarcity of movies that feature "weak" or narrow artificial intelligence, which is more representative of the current state of AI technology. While it might not make for a thrilling movie plot, exploring the limitations and biases of machine learning models could provide valuable views into the real-world challenges of AI development and implementation. For example, it is essential to compare the consequences of a potentially reliable 'traditional' machine learning model, which is trained on specific datasets to perform well-defined tasks, with the often opaque and unpredictable results generated by large language models like ChatGPT."

The research from Hammele shows that we discuss far too little about the small changes brought about by AI that eventually lead to greater results. For example, I wrote about the effects of generative AI on the freelancer industry.