Reporting from the 2018 Sundance Film Festival
Frankenstein AI, A Monster Made by Many
“Frankenstein AI: a monster made by many reimagines Shelley’s seminal work through the lens of a naive, emotional, and highly intelligent “life form” – an AI. Rather than taking a dystopian perspective on this rapidly emerging technology, Frankenstein AI brings participants into a narrativized world where the AI recognizes its need to better understand humans, in its desire to connect authentically with them.”
How do we define ourselves? What is it that makes us human? Do we understand our own humanity? How do we define AI? How do we create a comprehensible code on understanding humanity, or a dictionary on us, if you will? These questions became apparent throughout the Frankenstein AI experience.
When my name is called, I receive a card from the host with instructions on how to proceed. I text a number to receive a survey link, and on this survey, I identify the question that I find to be the most interesting. The questions are along the lines of this: what are your fears, why do humans kill one another, why do humans eat for fun, and what is the most human experience you’ve had?
Next, I respond to questions related to my sentiments on sharing personal stories with new acquaintances. Once this is complete, I am partnered with another in the experience based on the color of my card.
Finally, we enter the experience. It begins in a dark room with 6 tables and twelve chairs. Each table is rustic, but equipped with a digital surface. Here, we receive another set of questions that we read to one another. I am asked about a time I felt truly alone, and my partner is asked about a time he felt true connection to another individual.
When the timer rings, our table lights up. We use a little disc to highlight emotions and to match them with certain body parts. For example, there is joy, empathy, connection… to match with the eyes, heart, brain, hands, stomach, and mouth.
Now, we move to another room where a “doctor” stands to one side of a large screen that displays swirling lights. The AI perhaps? Yes, its visual representation.
After input from our earlier online survey, it is the AI’s turn to ask us questions. These are the following that it puts to us:
When is loneliness a good thing?
What do you do to make someone like you?
If you could have anything what would it be and why?
How do you think an out of body experience makes humans different from animals?
The answers we give are entered in as information for the AI to process.
The more we are pressed to contemplate these questions about humanity, we realize that our answers vary significantly.
How would you respond to these questions? Does this exercise explain to you what is humanity or do you have many more questions? How do we define ourselves? It’s not as easy as I thought.
This Frankenstein AI experience also reminds me of the 1995 movie Ghost in the Shell that I watched a few days ago. The film raises interesting questions about what it means to be human and how we can really prove our existence. Turning the tables, this story is told from the viewpoint of a cyborg who works alongside humans and a sentient being that evolves from a program. Do you think it’s possible to define humanity rather than “human?”
Back in the Frankenstein lab, the AI disappears to search for more answers on the web. We discuss its creation and its “personality” that is based on the Frankenstein story. The creators explain that the AI has been angry all week, as Frankenstein is a very angry story. However, after our input and the input of others who participated in the Frankenstein AI experience, its character may change.
Will the Frankenstein AI be a monster made by many? Or will it reflect another part of our human identity?
I recall the discussion with my partner less than 15 minutes before during which we contributed our emotional and conscious decisions on emotion to the creation of the AI. When we did this, we answered very personal questions to share with someone else in the experience. What is a true and unguarded response or a guarded response? Do you guard your emotion in unfamiliar surroundings? How does this affect the AI we create and how its program understands humanity?
All in all, it’s not necessary for an AI to understand us. An AI doesn’t care, it’s a program, but an AI trained on human emotion could provide some interesting insight into us.
After all, it’s us who create these AI who can exhibit such human “behavior,” and it’s the fears we have of humanity that is mirrored in our fear of AI. “Our fears of AI come from the fears we have of humanity, knowing that it’s us who control AI,” says one of the creators.
Beyond this, there are many other considerations such as the subjective interpretation of words, gestures, questions etc.
Then, how do we differentiate emotional mimicking v. emotional understanding? Think of the wink in “I, Robot”.
While it’s certain these questions have occupied a lot of people for some time already, even well before the application of AI came up, this discussion at Sundance is my first introduction to the idea that our interest in AI is really an interest in ourselves.
There is a lot of focus on technology itself and the effect it has, but what does this say about us?
–food for thought–