arc: An AI Co-Creation Experience

Every work of art causes the receiver to enter into a certain kind of relationship both with him who produced, or is producing, the art, and with all those who, simultaneously, previously, or subsequently, receive the same artistic impression.

A real work of art destroys, in the consciousness of the receiver, the separation between himself and the artist.

Leo Tolstoy, What is Art?

Think about the last time you were genuinely moved by a piece of art. Maybe it was a book, or a movie, or a painting in a museum. Maybe you were moved to tears, or to laughter, or to worship. What adjectives describe that experience? How did it affect your relationship with yourself, with another person, or with the world around you?

I was in Rome the first time I read Tolstoy's What is Art?, immersed in the world of Caravaggio and Bernini. But his ideas about art's ability to foster a relationship -- even an identity -- between creator and receiver have found resonance in my day to day life. They've been brought to mind by Bo Burnham's stand-up comedy special Make Happy, which examines the vulnerable relationship between performer and audience. They've been brought to mind by the fiction of David Foster Wallace, who portrays in intimate detail the despair and triumph of recovering addicts in a Boston-area half-way house. Art is a means of experiencing the experiences of other people with whom you share almost nothing in common besides your common humanity. As Tolstoy wrote, art is, "a means of union among men, joining them together in the same feelings, and indispensable for the life and progress toward well-being of individuals and of humanity."

But what if the artist whose work you're experiencing isn't a human at all? That's the question provoked by designer Jacob Elias' latest project: arc [1]. This experience invites you to co-create a short film with an actual, functioning AI. I'll leave it to Jacob to explain how this works; he's kindly written an exclusive introduction to arc, which you can read below.

Arc provokes questions about the nature of art and creativity, the sometimes-blurry lines between humans and machines, and role that social systems and technologies play in the production of human culture. If art truly "destroys, in the consciousness of the receiver, the separation between himself and the artist," how will artistic collaborations with AI influence our understanding of the self? What are the implications of inviting machines into the work of crafting the stories through which we make sense of our world? Can machines truly create something new, or are they bound to simply recombine existing elements? If the latter, is this really so different from human creativity?

Jacob's contribution is timely, given the recent conversations around OpenAI's GPT-3 AI and its ability to write original works of creative fiction [2]. You can read his introduction below or go straight to arc now by clicking here: https://arc.jelias.me. When you finish, please consider taking Jacob's survey at the bottom of this page to help guide future iterations of arc. I hope you enjoy!

Creator's Introduction

by Jacob Elias [3]

On Friday, June 12th, I launched my senior thesis, arc, a collaborative experience in film, AI, and the shape of stories.

arc is a generative installation where you can collaborate with AI to create your own films. You, the human collaborator, choose the emotional shape of the story. Then, arc selects and edits together clips from a library of existing films that correspond with the emotional arc you’ve prescribed. The result is  a completely unique story, co-created by you and an AI. 

I built arc to highlight society’s role in the creation of future systems, and to explore how our experiences, biases, and perspectives shape how technologies and systems affect our world. Moreover, in the context of our current world, art and humanity are ever more important—the arts allow us to connect emotions and stories to look at the world through a critical lens and challenge what we consider possible today. 

Users engage with arc to explore these complex conversations in three distinct layers. Through direct manipulation, users learn how their inputs directly contribute to and influence the final film that is created. Later, when prompted to create another film, users further reflect on how the guidance and parameters they provide directly contribute to the AI’s process and generated outputs. How does my direct interaction and manipulation of information and systems contribute to the generated outputs of AI? What is my role in these systems?

Diving deeper, upon learning how arc works, users explore the foundation of the system and how the AI is first built and trained. With each step in further unmasking the black box of the system, users are able to question how different training data, editing rules, and other foundational steps play a major role in the AI’s understanding of the world. How are these generative systems created? What data—and innate biases and perspectives—were part of the AI’s building and training? 

This project closes out with a space for people to further reflect on their opinions of AI and of collaborating with these technologies. The hope is that users will carry on these reflections, questioning how their biases and perspectives of the world may play a role in the technologies and systems of the future. What happens when we fail to address bias and perspective in complex systems? What happens when an AI’s training, or the building of any system, is not inclusive? How do our experiences, biases, and perspectives shape and guide the future of technology?

Finally, as an art installation, arc drives home an idea that AI does not exist in a vacuum. While the artist (user) may not consider themself a filmmaker, their experiences and perspective is the central input to creating a film, and so both the AI and user become the artist. What is art with technology? Who is the artist? What does a future of collaboration with these systems look like?

I cannot say I have the answers to these questions, but reflecting on my experiences with these systems has been extremely insightful. I know that only through diverse and inclusive approaches will we begin to design systems and technology that benefit everyone. These conversations, reflections, and introspective moments—challenging our own work, ideas, and understanding—are key to learning and creating better.

Click here to create a film with arc: https://arc.jelias.me

Then, consider returning to participate in the survey below to help shape future iterations of this experience.

Notes

[1] Jacob Elias is an Interaction Designer based in the Bay Area. His previous design work includes collaborations with Google, IDEO, and Short of the Week. You can contact him and keep up with his work through his website: https://jelias.me

[2] Learn more about GPT-3 here: https://www.economist.com/science-and-technology/2020/08/08/a-new-ai-language-model-generates-poetry-and-prose; for some examples of creative fiction generated by GPT-3, see here: https://www.gwern.net/GPT-3

[3] See [1]