PODCAST + Discussion: A.I. and Creativity with Justin Barber

The second "I Watch So You Don't Have To" podcast episode has been released!

I recommend you listen to the episode and return here to dive deeper into the storytelling and creative possibilities of artificial intelligence.

---

Are We Doomed?

In the midst of our current Armageddon-like pandemic reality, it's easy to feel as if the end is nigh. Or, at the very least, that massive change is on the horizon (a horizon that is moving a bit to close for comfort).

I do not say this to induce more panic, nor do I aim to lull us into ignorance, but I would like to foster a conversation around the big changes in A.I. that could bring us hope (rather than trembling) at what lies ahead...

Movies and TV shows like Ex Machina (2014), Westworld (2016-present), I, Robot (2004), Her (2013), Black Mirror (2011-present), Transcendence (2014), and countless others have been delivering robots-take-over-the-world scenarios to our living rooms and big screens for decades, but are our A.I.-based fears warranted? The primary fear we humans seem to cling to is the concern that machine intelligence will be able to mimic, outpace, and then replace human intelligence. But does creativity give us an edge? Can robots create original works of art, evoke emotional responses, or tell stories that we connect to?

How Do We Define Storytelling?

Answering these questions involves reflecting on what we mean by "creativity." I am particularly curious about storytelling, and the ways in which we have continued to produce and attach ourselves to on-screen stories (movies/TV). Storytelling and creativity seem to be innate elements of "what it means to be human," but we would have to define creativity in order to program a computer to do it. Translating creativity into an algorithm is beyond the scope of this post (and far beyond the scope of my expertise), but luckily by podcast conversation partner provides us with a useful starting point. In this week's podcast I had the privilege of discussing all of this with Justin Barber, A.I. Engineer at the Iliff A.I. Institute and Lead Data Scientist at the Iliff School of Theology.

Justin looks to Mihaly Csiksgentmihalyi to help us consider the creative process. According to Csiksgentmihalyi, three things that need to exist for creativity to be present: symbolic rules, novelty, and expertise. A system of symbolic rules forms a space in which we can use universal language, images, metaphors, concepts, etc. to communicate ideas. The novelty of generating something we’ve never seen before is the magic ingredient that captures our attention, sparks our imagination, and keeps our curiosity thirsty for more. And finally, whatever is created needs to be validated by a experts in the field — is this a good painting, film, song, poem, garment, compared to our understanding of other paintings, films, songs, poems, and garments?

Source: Analytics Vidhya (July 29, 2019)

Advanced NLP (Natural Language Processing) programs like Open AI's GPT2 language model can create sentences by predicting each word in turn (image above), but could it predict the next plot point? There are some hilarious examples of A.I.-generated screenplays, which mostly confirm that machines will not be replacing human storytellers anytime soon (image below). But could this change?

Source: Tweet by Keaton Patti (@KeatonPatti), Aug 13, 2019.

Creativity and Authenticity

There is an underlying question of authenticity at stake here. A.I. is an Other we do not fully understand and are, most likely, suspicious of, so it is difficult to truly parse out what may or may not be authentic. Perhaps, instead of judging A.I. creations based on human notions of creativity, we might ask: what kinds of stories would an A.I. tell?

This encounter with the Other is displayed quite gracefully in this Vogue-produced video in which the young, Grammy Award-winning musician Billie Eilish is interviewed by a robot (programmed by Nicole He). The robot asks questions and even presents an original song inspired by Eilish's music.

Clockwise from top: the robot interviewer asks a question; the camera component (the "face" of the robot); Eilish comments on a song created by the robot; the robot's song lyrics scroll across the screen.

Throughout the interview, moments of surprise flash across Eilish's face, revealing her unexpected openness as she encounters this robotic Other. We are so used to human-to-human interactions that any deviation from the norm feels jarring. We seem in need of translation, interpretation, and criticism -- something on-screen stories have had since early films flickered into cinematic life. Just like Amy Adam's character in Arrival (2016), we need experts (film, television, and other art critics) to help us make sense of A.I.-generated works.

In other words, just because something is new or Other does not place it beyond comprehension. We simply need artistic experts (alongside computer programmers) to become our new interpreters of A.I. creativity.

Not unlike the concerns expressed by Walter Benjamin, who believed that mechanically reproduced art (e.g., film) creates a lack of authenticity and unoriginality, we might wonder how a machine programmed to create something is truly responsible for the creation itself? Does a work created by an intelligent machine authentically belong to the machine...or to the programmer...or somewhere in between? This hybrid in-between space forces us to consider creative collaboration, rather than ownership, as the location of A.I. creativity.

Robots: They're Just Like Us!

So what does this collaboration look like? In the age of streaming services and declining attention spans, A.I. can help film and TV storytellers embrace the changes already sweeping the industry. For example, current storytelling is increasingly in short form (e.g., the emergence of Quibi and skyrocketing popularity of TikTok) and A.I. could help with this move toward short “scenes” rather than stubbornly clinging to cinema and long-form serialized television as the standard for real or authentic creative storytelling. We needn't fear the fate of classic cinema — there will always be those of us who crave the moviegoing experience — but rather, we should welcome these innovative nuggets of storytelling. As Justin puts it, this kind of storytelling "adds rather than detracting."

Justin also reminds us that A.I. "bring[s] something to the table that we don’t have, inherently," such as comprehensive data intake. While humans think with "facts…and classical logic," A.I. often utilizes "probabilistic" thinking. This is why Billie Eilish was so taken with her robot-interviewer; the novelty of the questions was unexpected and refreshing. Yes, a human being programmed the robot, but the randomness of the questions were generated and presented by the A.I.

Imagine a creative future where artists and writers incorporate machines into the creative process: random prompts could awaken fresh ideas; databases of existing screenplays could be utilized to understand plot formulas (in order to embrace or avoid them); casting and production data could be collected to help us survey and combat the lack of representation and inclusion in the entertainment industry (something that USC's Annenberg Inclusion Initiative is currently spearheading).

I echo Justin's hope that "human beings will always partner with A.I." in other to see things in a new way, rather than fearing the algorithmic Other.

Conclusion: Storytelling, A.I., Meaning-Making, and Religion

This is all quite interesting, but is it feasible? In short, not right now, nor in the immediate future — we are limited by the technology, as it stands today.

As noted by artist and Partner Director of Ethics and Society for Cloud and AI at Microsoft, Mira Lane, machines still have difficulty creating the "organic-ness" associated with human creativity. They are able to mimic, but not quite able to harness what we might call artistic authenticity. But if today we are asking these kinds of questions, then there is the distinct possibility (and, perhaps, probability) that we are moving toward this future reality. A reality of A.I. and human creative collaborators may be at hand.

But we still have a long road of programming and creative thinking ahead of us. Along with symbols, novelty, and expertise, Csiksgentmihalyi also believes fluidity and fluid states are necessary for creativity. Can intelligent machines embrace this fluidity and "organically" produce the symbolic and novel elements of creativity? How would we program this sort of creative approach?

The more important question might be: Is the goal of creative machines to mimic human creativity, or is this a fundamentally different experiment that requires us to be open to the Other? What kind of story might a machine tell, and would we enjoy it? Would we even understand it?

And the Oscar goes to...Hal?

Source: 2001: A Space Odyssey (1968)