AI Perfection and the Heart of Filmmaking

Previously I wrote about tolerable imperfection. Of course, all the happy accidents can be mitigated with unlimited resources and money, leaving only intention, but that wouldn’t be as exciting. And in my experience, when unexpected things happen, that’s often when the most magical outcomes occur. This is one reason I’m not overly worried about AI in media production, I don't see innovative serendipity in generative AI being overly satisfying, it's more like generative mistakes.

AI is only one factor in a broader realignment and evolution of moving-picture content creation. For the first time (at least to my knowledge), every level of content production has been affected in largely similar ways.

I think there is however, a distinction to be made. A lot of the discussion surrounding the impacts of AI on content production conflates effects across sectors and niches that will not be impacted in the same way.

For documentary films and very low-budget independent narrative work, AI tools are a massively beneficial resource that will positively impact the level and polish of storytelling possible on very restrictive budgets. Those benefits certainly outweigh the negatives in virtually all scenarios imaginable. But in other areas—from the low end of corporate communications through social media channels—the disruption will be quite significant. Why hire and pay for a video shooter to capture something you can ask AI to make in seconds?

And on the other end, in the space where the traditional film-studio model operates, the emergence of AI is just one of many factors working in concert to make the disruption to the overall business model more profound.

Make no mistake: what we are experiencing isn’t just a cycle of change in the film and video business; it’s connected to the broader social, political, and economic uncertainty that permeates every aspect of our lives.

Take, for example, corporate communications, which for a significant number of production workers has been an income/revenue backstop to fill in the gaps between narrative project work, even if most don’t like to talk about taking on that kind of work in slower periods. When a company isn’t certain what to communicate and cannot reasonably predict the consequences of a course of action, it simply doesn’t allocate budget, content just isn’t made and that affects everyone because it ripples and displaces talent at various levels as more and more people are willing to take on less desirable work for their skill level.

The money waits until conditions improve or when there are significant enough financial pressures that force a decision. But in the meantime, that fill-in-the-gap work for so many, simply exacerbates the uncertainty and volatility of also not seeing the same levels of opportunity in the other areas of production work.

There are areas of production and post where AI will make a massive difference—what can be achieved at a given budget level—and it will result in fewer people being engaged to work on projects that ten years ago would have seemed impossible to accomplish.

It can make rotoscoping or chroma keying faster and better with less time and energy inputs. But it can only iterate on what has been done; it cannot create entirely new concepts or anticipate happy accidents. It doesn’t have humanity or emotion; it has no soul and cannot anticipate how the images, pacing, color, or framing decisions might resonate with the viewer.

I believe that AI will open new possibilities to make real what our imaginations conjure and lift the limitations on what we can dream of due to budget constraints, but it’s not going to do the entire job of filmmaking. Someone will need to create a concept, nurture an idea, control how a story is told, and make decisions about picture and sound that serve the story and are entirely subjective, subservient to how they make the viewer feel. AI can only approximate elements of that; it cannot feel for us.

There is a lot of conversation around the state of the industry, the impact of AI, shrinking budgets, and fewer projects (seemingly) being green-lit. But what are we talking about? Feature films, series, documentaries? Or even commercials, regional, national? Corporate communications? Safety and training videos, or even the most fleeting of content, content strictly for social?

In a way, we can broadly define two primary types of content: low-value content that has a very limited life span—like the vast majority of social content—and higher-value evergreen content that has relevance and intrinsic value either for a long time or in perpetuity.

Everyone has experienced a contraction in demand for content of all flavors and purposes, and myriad discussions are occurring. I think those conversations get muddy with the conflation of distinctly different markets and types of content. We should endeavor to maintain context. A great deal of the discussion has related to “filmmaking”—the scripted-drama content that most people think of when considering the implications of AI and other industry disruptions.

But there are vastly more people currently working in corporate communications, commercials, documentary, and industrial video work than there are in scripted content production. These are typically one-person operations or very small crew teams, and due to the accessibility of technology, are producing content at a quality level unimaginable ten or twenty years ago.

The current challenges in the industry will result in fewer people being able to make a living creating a multitude of content that has traditionally required skill and knowledge. For a great deal of that content, specifically for social media, I don’t see a future where an individual can make a living as a service provider alone without some deeper value-added to the equation.

In my production work there are times when we need to scale up the crew to handle potential challenges we can predict could become a problem. But I think scaling has more to do with how fast we need to move, or what safety concerns there are for all involved. Rather than simply adding bodies for positions because that’s what has traditionally been done. That has been the trend for a long time now and will continue.

AI cannot provide heart, and especially for documentary storytelling, it’s simply not possible. Because the truth is, AI isn’t intelligent; it’s simply good at iterating by drawing on lots of information to create something narrowly defined. If something is created by AI and can be recreated with a prompt, where is the value for the original creator of the idea? There must be something more intangible that breathes life into the work, something human.

Don’t get me wrong: AI has, will, and will continue to make mundane and repetitive tasks basically automatic. And for someone like me who has spent hours rotoscoping the individual strings of a tennis racket out of an image of Serena Williams for a motion-graphics composition, I welcome the day that job is no longer one I’ll be asked to do.

In the meantime, we should try to maintain some perspective as it relates to the wide variety of work we are often asked to work on. There are just as many opportunities and benefits to the changes we are experiencing as there are negatives and potential job losses. The upheaval isn’t new; the cause of the upheaval is. We just need to figure out how to adapt our individual skill sets to play slightly different roles in our own stories.