AI | AI: Making of Reality

Ask the AI: How old is Robert de Niro (center) in “The Irishman”? Photo: imago/Everett Collection Sylvester Stallone takes a running jump into the gorge, pursued by cameras and the director's eyes, held by safety ropes and his muscular upper arms. Can be seen on Tiktok. The actor holds on to the oncoming plane until it prepares to land. The film set applauds, the stunt seems to have succeeded. But something is wrong with the scenario, the background, the movements. The same applies to the miniature city surrounded by cameras, which is either destroyed by tsunamis or tornadoes for a disaster film, but neither water nor storm behave physically correctly.

And the lovers on board the “Titanic” also receive instructions from a wooden James Cameron, surrounded by an unnaturally laughing crew. Rose and Jack, unequal in their social status and especially in quality to their Shakespearean role model, move so unreally that they can hardly be trained actors. And actually: they are not. They are all AI-generated scenes that are presented on the relevant platforms as “making-of” or “behind the scenes”.

Such and similar scenes have become the latest AI trend, often with several million views. Now there are at least as many trends in the AI ​​world as there are people willing to consume them. With a touch of media literacy you can see that most of it is complete nonsense: disproportionate, cheating and revenge-taking fruits, singing and dancing pets and a Putin-kissing Zelensky. They are clichés or fantasies, often underpinned by racist and political motives.

But things are different when it comes to the supposed look behind the production scenes. Sure, some of it is just as stupid as the love affair between eggplant and strawberry, especially Stallone, who jumps into a volcano with a bowl of corn only to make it rain popcorn shortly afterwards. Instead, it's the grounded and slow videos that are dangerous. The laughing dinosaur on the film set is immediately recognizable as a fake, while things become more difficult with the calm cameramen who are absorbed in their work. Although these can still be recognized as generated Pappenheimers, the speed of developments in AI and the like shows us that it will not stay that way.

This may be good for a smile, even believable for some, but it should be another cause for concern.

Where it says “Behind the scenes” and “Making-of”, the real thing should be shown. When the camera and director's chair are visible, the audience must at least accept the break to the production level. Of course, you can also play with it, as the production can also be part of the work of art and the creation can be interpreted dramaturgically. Films like “One Cut of the Dead” and “Shadow of the Vampire” don’t work without interweaving the played and rotating levels. For us, the audience trained in pop culture, this is and must be transparent at a certain point. But this is exactly where the artificial generation comes in and produces a false image of supposedly real productions.

The result is generated insights behind the equally false scenes of the “Harry Potter” and “Jurassic Park” films. What never happened is staged as a making-of (a special highlight: the film crew creates the steam from the Hogwarts Express with water pipes). This may be good for a smile, even believable for some, but the general affect should be one of concern. The literary world already shows how much AI has created uncertainty and an atmosphere of mistrust there.

There it is the subconscious “kick” that many people feel when unconsciously reading artificially generated literature, triggered by a gut feeling fueled by inappropriate metaphors or repetitive formulation patterns. In his essay “Who cares who doesn’t speak?” the cultural scientist Johannes Franzen aptly summarizes the paranoia about the artificial in the literary world: “The rise of generative AI has created new possibilities for authorial fraud and has thus given rise to new fears and concerns that have already significantly shaped the literary field.” The result is, in addition to an increasing distrust of authors, also the devaluation of institutions such as criticism, which until recently boasted of recognizing human texts as such can.

As it gradually becomes more difficult for AI models to recognize texts as such, the same also applies to static and moving images. In February, the “Heute Journal” fell for a clip from the now-defunct video AI app “Sora,” which showed ICE officers doing their illegal work (even though their illegal deportations actually happen every day). In combination with the deceptive glimpses behind fake scenes, they show how easy it has become for software to present us with false realities.

The emergence of computer graphics in the 1980s caused lively discussions in romanticized Hollywood and other film studios about the real and the synthetic, the credibility of the medium and the role of the actors in a technological environment. The digital resurrection and rejuvenation of the deceased and the aged – see Peter Cushing in “Rogue One: A Star Wars Story” and Robert De Niro in “The Irishman” – followed the same lines. Until now, complex computer effects required staff with the appropriate know-how and de-aging still covers up the wrinkles on a human face. But the increasing use of generative technology further attacks human achievements in the medium of film and - the false making-ofs make it clear - further weakens the divisions into "real" and "false".

At this year's Cannes Film Festival, Tilda Swinton wanted more "chaotic, adventurous experiences" and less formulaicness in response to the invasion of AI into the film industry. But a look at the German mainstream cinema landscape makes you depressed and, at best, pessimistic. Perhaps the generative advance also leads back to the medium that created the film and has not yet been able to replace it: theater. But that would probably be too much cultural optimism.

With us, no one has to pay to be informed. Sounds like a bad business model? It is too. But independent journalism and participation for everyone are more important to us than profit. Journalism still costs money: research, printing and all the people who do it. That's why we depend on you. We need 300,000 euros by the end of the year.

Donate once or regularly, every amount counts. So that nd remains.