In late March, the innovative artificial intelligence startup Runway hosted its first ever AI film festival at the Alamo Drafthouse theatre in downtown San Francisco. Runway had been well-known prior to this for developing Stable Diffusion, an AI text-to-image tool which had been a huge success in 2022. In February of the same year, they had released a video-editing tool, prompting many filmmakers to come up with their own creative works of art. 10 of these videos were selected to be shown at the film festival.
Most of the videos demonstrated the power of technology as opposed to traditional storytelling. Several of them were surreal and one of them was even macabre. The last movie shown was particularly spine-tingling. It was titled Expanded Childhood and instead of being made up of video footage, it was composed of photographs which had been altered by Runway’s AI tool, DALL-E. The narrator was barely audible, so much so that it felt like it was part of the surreal atmosphere this film was generating.
Expanded Childhood was the work of Sam Lawton, a 21 year old film student from Nebraska. He used OpenAI’s DALL-E to create the images used in the film. He had taken childhood photos of himself, and then used AI to manipulate the images in various ways. He asked the AI to complete the images, fill in the edges with more items or people, or to change the images to give them a different look. His father’s reactions were also recorded to add to the film. In awe, he said things like; “No, that’s not our house. Wow — wait a minute. That’s our house. Something’s wrong. I don’t know what that is. Do I just not remember it?”
With the rise of AI powerful tools, questions are being asked such as ‘Where does real memory end and generative AI begin?’. Google and Adobe have both released AI-based editing tools which completely change the context of images and may even lead to confusion over the truth of a situation. Google has the Magic Eraser program, and now is testing out Magic Editor which can remove objects from images, and Adobe has developed Firefly and the Generative Fill feature which can insert items into images simply through the user entering a text-based prompt.
The Runway startup has demonstrated how AI can be used in creative and powerful ways to manipulate images and imagination. This has opened the doors to new possibilities and often blurs the lines between truth, memory and photoshop. While some of these tools can be seen as helpful assists, they are also capable of drastically changing how we view the world, challenging the way we think and remember our pasts.
