A paper by Eleonara Rosati titled The future of the movie industry in the wake of generative AI: A perspective under EU and UK copyright law states the following:
…some have stressed the opportunities presented by the implementation of AI, including by advancing claims, like those made by AI video studio The Dor Brothers that at AI tools ‘are actually a purer form of expression, offering the most direct link between the artist’s brain and the end result, without the compromises required in large productions or the constraints that come with complex shoots’
The quote by The Dor Brothers raises a question I imagine many creators ask all the time—why use generative artificial intelligence (GAI) to produce anything? The answers will vary depending on the medium of expression—from the sculptor who says “never” to the audio-visual producer who says “all the time”—because beyond the legal issues triggered by GAI, the technology reframes the question of what it means to create works of expression in the first place. And this includes the question as to whether removing “constraints” is either conducive or harmful to the creative process.
Although motion picture production entails more non-creative constraints (e.g., large investments and complex logistics) than all other media, I would caution that even in filmmaking, constraints are generative of creativity. In the same way that working around copyright constraints tends to produce new creative expression, this is also true of the limitations inherent to each medium. Moreover, the idea that an artist does not want to confront the constraints of her chosen medium is misguided, and the passion to confront those challenges is not a matter of mere nostalgia.
I get what the Dor Brothers are saying, of course. The AV producer can go from script to screen without any of the costly and cumbersome production work that will frustrate, if not substantially alter, the original vision. Screenplay material becomes prompts, and the GAI outputs the AV material without the need for cameras, actors, sets, etc. Still, the extent to which the outputs more “purely” represent the mental conception in the “artist’s brain” is both a question of copyrightability and artistic integrity. How much control the AV prompter has over the resulting material will determine the extent to which he owns the rights in that material, but even with extensive control, the “purity” of the expression is not necessarily preserved by the removal of constraints.
Notwithstanding many useful applications of AI, including for various aspects of artistic work, all the talk about “democratizing” creative expression (i.e., without developing skills in various crafts) reprises that question Why? for many artists. If you don’t enjoy dealing with the constraints of clay, paint, words, light, sounds, etc., then you probably don’t really like the process of creative expression. Again, that’s not just luddite’s nostalgia. Creative expression (art) results when the unique, imperfect human confronts, learns from, and eventually masters the constraints of a chosen medium. As my friend Sandra Aistars, copyright professor and, recently, a fine art student, writes about the distinction between AI “training” and human learning:
… instead of predicting “what comes next,” artists studying masterworks are taught to unlock “how” the original artist has conveyed what is foundational to an image’s storytelling. This requires patience, humility and empathy on the part of the artist asking to learn. But it ends in developing one’s own aesthetic judgment and voice.
Aistars describes engaging with the constraints of visual artmaking by retracing the steps of masters in order to discover her own aesthetic. The process is physical, intellectual, and emotional at the same time, and most artists would ask why a creator would want to avoid engaging with the medium in this way. It is the act of confrontation and the artist’s unique mode of problem solving where the meaningful act of creating occurs for the individual.
Using GAI as a cheap or free assistant to write a boilerplate email or report makes sense, but the hyped-up marketing of these products, challenging users to push AI to “write poems or novels” is asking people to fool themselves. You might have a brilliant idea for a premise, but if you don’t want to grapple with the constraints of writing, you’re not a novelist any more than you’re the “boyfriend” of an AI companion.
Turning back to the Dor Bros.’ comment, because motion picture production entails thousands of constraints that are not necessarily generative of creativity, their point has some merit in certain applications of the medium. Specifically, a lot of their work appears to be commercial advertising at this time, and the utilitarian nature of marketing material, combined with the attraction of low-cost, fast-turnaround production cannot be ignored. Nevertheless, I would caution against the idea of a “pure” link between an artist’s “mental conception” and the end result by means of removing constraints.
Motion picture production still entails many constraints that are generative of creative expression. Just as Aistars chooses to wrestle with the possibilities and limitations of a particular pencil in her hand, the filmmaker has a complex set of “tools” that include the constraints of physical space, light, camera and lens characteristics, performers, writing, time, which must be confronted to find the film’s unique voice. And as any film student can tell you, working around constraints has often resulted in moments considered to be works of cinematic genius.
Naturally, GAI is already used to reduce or eliminate certain drudgeries in creative production, and although this also implies reducing or eliminating various jobs, that is a separate matter from the philosophical premise to which this post responds. In general, I am skeptical that a seamless, constraint-free transition from mental conception to creative expression is desirable, even if it is achievable. Constraints define the various artistic media, and it seems more likely that expression through GAI will evolve as its own medium with its own constraints. Otherwise, if GAI’s only purpose is to synthetically displace the creative process in all media, the results will likely be as bloodless as the computers that made them.
Photo by: Ponsulak
