Photo: HFF München / Robert Pupeter
Sylvia Rothe: How AI is changing film production
Prof. Dr. Sylvia Rothe is an AI professor at the University of Television and Film in Munich. She wants to teach her students in a practical way how AI can be used in films.
Ms. Rothe, will AIs rather than people be awarded the Palme d'Or in Cannes in the future?
Sylvia Rothe: I hope not and, to be honest, I don't believe it either. Behind every AI application there are lots of people who have thought about what tasks the AI should solve. Ultimately, they are the ones who decide whether the results of the AI are useful for their purposes or not. For example, AI is a helpful tool for visualizing people's ideas or implementing texts. However, the original idea is the human one.
Are scripts already being created entirely with AI?
Rothe: Technically, this has been possible for quite some time. The first experiments were carried out a few years ago, but back then AI only generated nonsense. Today, AI is more advanced, but screenwriters don't let it write their entire scripts. They use it more as a sparring partner: they enter a topic and let themselves be inspired by the AI's ideas. The result is a kind of conversation from which loglines, outlines or characters emerge. Normally, you would ask a colleague for advice who would give you tips - but these people usually operate in a similar bubble to the one you are in. The wonderful thing about AI is that it pools the experiences of many people. As a result, the AI spits out ideas that you would never have thought of otherwise.
„In documentaries, the problem often arises that you have to separate the voice of the protagonist from the background or other voices. At the moment, this is still very time-consuming [...]. AI can make this process more efficient.”
AI has a bias – does this influence the stories of the films?
Rothe: In a way, AI holds a mirror up to us: It accesses the data on the internet that we ourselves have created and draws conclusions from it, in which we in turn can recognize a bias. I also see this as an opportunity, because only if we recognize the bias can we actively change it. Without the ability to analyze this mass of data, we might not even notice it. The providers and users of AI models are aware of this and are trying to counteract it. That's why I believe that no films will be made that adopt this kind of bias one-to-one.
Where is AI already being used in film other than in screenwriting?
Rothe: It's used in sound, for example. In documentaries, there is often the problem of having to separate the protagonist's voice from the background or other voices. At the moment, this is still very time-consuming and sometimes simply not feasible. AI can make this process more efficient. Material sorting for post-production is also made easier with AI tools: AI can find certain camera angles of a protagonist or specific dialog easier and faster than a human.
In future, complicated camera movements will no longer have to be carried out using tracks or cable cars, but can be generated by the AI. The same applies to expensive real-life explosions. In production, we can also use it for feasibility analyses of a shoot, financial forecasts or audience analysis. In the future, we will see many exciting developments, perhaps even personalized films. A movie that ends happily for one person and with a dramatic finale for another.