The AI short film Mnemonade (2024, Drama, MetaPuppet), winner of the inaugural Culver Cup for generative AI, is the the product of a talented filmmaker and storytelling, with good structure, believable dialog, and a twist that will make you hug your mom.

I’m not afraid of AI, particularly AI video. In the last couple of years, some artists and writers have demonized AI as theft or misappropriation, an attitude rooted in a misunderstanding of the technology or plain old willful ignorance. They also claim AI will take their jobs. Though genuine questions remain about the use of images and text for training, the history of copyright points to a general acceptance of the “fair use” of copyrighted works in other media. Some jobs will be lost of course, but history also shows many more will be created that require similar—and different—skills. And honestly, generative AI has already established itself within a significant portion of the artistic community; resisting artists had better adapt or they’ll be left behind. This is the nature of technological progress.

Mnemonade (2024, Drama, MetaPuppet) is the the product of a talented filmmaker and storyteller, with good structure, believable dialog, and a twist that will make you hug your mom.

A powerful template for the future

As the winner of the inaugural Culver Cup for AI film excellence, Mnemonade offers a powerful template for the future of narrative AI media. Its strength is not in its brilliant use of AI video and audio technology, but its compelling story. An elderly woman named Judy arises, mummy-like, out of the sands of time into a 1950s diner, where she meets Frank, her future husband. The 4:56 film shows her moving through marriage, children, and old age, and finally into an all-too-familiar present. Though the film is fundamentally a drama, it incorporates elements of horror that illustrates what it’s like to be trapped in your own memories.

Watch Mnemonade on YouTube.

Technically, the film has few flaws. The filmmaker, MetaPuppet, describes himself as an “Emmy-award winning video editor, writer, generative content producer, and consultant.” However, he says little about his tools and techniques. AI artists and filmmakers should share information on the tools they use. However, it’s clear he’s used the latest, most advanced generative AI and image manipulation tools. Despite the attention to details, such as the rich, almost leathery texture of Judy’s skin, the facial expressions for extreme emotions often appear exaggerated, almost rubbery, a common problem in AI face generation. The notorious “six-finger” rendering artifact shows up in a six-tined fork (most have three or four), though it’s only visible for a second or two. And text, by and large, is unintelligible.

Leaps in AI technology

Film history is punctuated with leaps in technology, which forever change the nature of visual storytelling. The breakout of generative artificial intelligence into the mainstream with the release of ChatGPT in 2022 is only the latest advance, and related tools have already changed filmmaking forever.

As with all art, there’s a ton of crap out there. AI film operates today in a no-rules, no-standards landscape. As a screenwriter, short story writer, novelist, and journalist who’s loved movies since my aunt took me to see Snow White and the Seven Dwarfs at the Capitol Theater in Yakima, Wash., I want to highlight the best in this new genre, and offer unflinching criticism of what’s gone wrong.

The end of gatekeepers?

Like many digital technologies, AI democratizes content creation. When I started my writing career, a limited number of gatekeepers decided what novels, magazine articles, poems, and so on could be published. Today, individual authors can publish on their own, and sink or swim in the stormy ocean of content. A few of the legacy gatekeepers still exist, and they have the resources to stay afloat, but the content sea is now full of boats large and small, which is probably a good thing in the long run. Filmmaking is rapidly approaching a similar outcome, made possible in part by generative AI. All it takes is an internet connection, a browser, a late-model laptop, and some elbow grease and patience.

Image: Still from Mnemonade, by MetaPuppet

Are you using generative AI for stills or video? Tell me about it in the comments.