Frankenstein AI Explores What it Means to Be Human

Artificial intelligence, one of the most promising technological developments of the past decade, has had a slow start finding wide use cases. Right now, it’s primarily being used by marketers and the military. But is that all it can do?

“The consciousness of AI up until a year ago was controlled by Hollywood and Silicon Valley,” says Rachel Ginsberg, one of the artists who worked on the design research project Frankenstein AI: a monster made by many, which debuted last month at the Sundance Film Festival. “It had a dystopian vibe.” Think Skynet or Siri — not exactly an inclusive vision of the future. “Can people even have a speculative conversation about AI as a tool for humanity, as opposed to war? We can’t,” argues Ginsberg.

That’s why she collaborated with Nick Fortugno and Lance Weiler to create their exhibit, which playfully introduces a level of democratic participation into the public narrative around AI, through the story of one of history’s most famous artificial intelligences: Frankenstein’s monster. Reimagining the monster as a machine-learning algorithm that takes emotional data as input from the participants and teaches itself how to feel human, Ginsberg hopes to create “a framework that makes a hard subject accessible, so people can learn and collaborate around it.”

AI plays two roles in the project. It’s a character in the narrative of this story, but there’s also a machine-learning algorithm running sequence-to-sequence processing. “We’re harvesting their hopes and fears,” says Ginsberg. The AI processes that into what she calls an “immersive experiential context,” and reflects it back to the participant, giving them the opportunity to respond in kind. “There is a consistent interplay between the AI gathering input, putting out output, and then people reacting.”

This multichannel activity works at the level of meta-narrative, too. “On the one hand, we’re exploring the story, and how the narrative can influence the trajectory of AI, and how we’re bringing it into the fold or not,” says Ginsberg. “Will we let the marketers and military determine the future of AI? How can we use AI as a new form for storytelling? How can it be used to augment storytelling in a way that’s boundary-pushing?”

The story of Frankenstein’s monster is an excellent way to push that boundary. The 200-year-old novel written at the dawn of the industrial era still has fresh insights to share about the protean nature of humanity as we begin to transition from a post-industrial to a potentially post-human age. Are our creations, as expressions of our capacities, in some way human themselves? Or do they have their own natures? Do they change what we are? “What we want to do is think about what it means to be human, by using AI as an analogy to have that conversation,” says Ginsberg. “How do we get people to get together, use that interaction to create new types of culture and create new types of work or solutions or futures, whatever that output? What opportunities are there to explore the function of story beyond just using it for entertainment?”

Ginsberg wants the audience of this immersive work to be able to see in it a trace of themselves. Making sure that kind of input/output between machine and people is evident to the people going through the experience is a main goal, as is serendipity management—“people bumping into each other in interesting ways.” But ultimately, the point is to grant people agency, over the Frankenstein narrative and over the emerging narrative around AI itself. Giving “collaborators agency to influence a story themselves, to really give people a number of different ways to act”—that’s what the Frankenstein AI teaches us being human is all about.