Multimodal AI, as I can imagine it this decade, will significantly disrupt the Hollywood industry.
From the outside, Hollywood appears to be two things - a financial commercial entity as well as a tight, regional network of professionals with expertise in filmmaking. Localized in the central region of Los Angeles, California, Hollywood has had so many generations of filmmaking and television experience, it has had an unparalleled influence on modern day global art and culture.
One of the most surprising things about Hollywood is the diversity of professional services available in just a few city blocks. There are VFX experts, a whole pool of the world’s most talented actors, but even little things like services which let you rent prop money. There are so many talented people who have expertise in pre-production, set design, casting, stunt direction and safety, lighting, accounting and finance, legal, production scheduling and more … all in the service of filmmaking. While yes, you can find these kinds of roles in filmmaking hotspots around the world, the concentration and intergenerationality of this talent is specifically unique to Hollywood.
It’s not only a pool of highly concentrated talent, but also a collection of unionized workers. These unions protect the interests of these workers and establish guidelines for their job especially around massive mega-budget hollywood blockbusters. Sometimes, these rules are arbitrary and create too much red tape, and other times they really do protect the interests of the workers who would otherwise be exploited for the creative work they love doing.
At the same time, Hollywood is also a commercial financial capital. Studio executives and producers work together to pick stories and teams which can deliver disproportional investment returns in the global box office. Typically, films are viewed as highly speculative investments to most traditional investors, but Hollywood, over many generations, may have developed a better intuition around which films will be commercially successful. I don’t want to give Hollywood financiers too much credit here, I wonder how many great films they haven’t funded and also what have they been funding lately? I wouldn’t necessarily say they are taking “risks” with their money.
Like most industries, Hollywood is constantly evolving. Set design changes every ten years or so. New technologies come about pushing the boundaries of filmmaking. In a short time, we’ve seen hollywood go from blue screen, to green screens, to virtual production sets. From film to digital. Technology has changed film distribution as well. Netflix, Apple TV, YouTube … these are all platforms which have changed the economics of the film business forever.
Part of the reason I think Hollywood is so receptive to change in many ways, is because I think they are always in service of three main things. The first being the story they are trying to tell, followed by the audience they are telling it to, and finally their own budget and profitability limitations. If technology can help them achieve these things better, like any other industry, I think they will pursue what’s in their selfish best interest. At the same time, I don’t think audiences care that much about how the sausage is made, whether it was filmed in real life or generated through AI (as long as it’s good and not distracting from their overall experience). This is an important thing to keep in mind.
The truth is, when I first saw DALL-E, and it’s ability to generate such impressive images from nothing, I was reminded of the cartoon that kickstarted Disney, called Steamboat Willie released in 1928.
Unfortunately, because Big Mouse Incorporated has fought to extend the copyright on this one … I can’t show you much here on Substack. But the point I’m making is that after DALL-E, I think the tech industry has experienced its own, “Mickey Mouse Moment”.
A moment which captures the imagination of technologists and creatives alike. Something which sets the bar but also the industrial commercial direction of what’s to come.
I for one, immediately saw the opportunity to generate entire movies, animation sequences, and visual works without the need for a large team or multi million dollar blockbuster budget.
A decade from now, imagine anyone typing in an advanced movie blockbuster script like Tenet and having AI completely generate many versions of it for you completely with all the bells and whistles. I’m talking IMAX format, world class music with full orchestra, highly customized beautiful computer generated human actors, and of course … uncannily good life-like special effects.
Keep in mind, Tenet is no ordinary film to have AI generate for you, at one point in the movie, the main characters crash a plane literally through an airplane hanger (which Christopher Nolan actually filmed in real life):
… but imagine if AI could generate movies like this for us, sparing us from having to spend 205 million USD in the real world like Christopher Nolan had to.
I’m calling this phenomenon of multimodal AI film and video making, Silicon Hollywood. I’ve even made up a definition:
Silicon Hollywood is a new approach to creating industrial quality film and videos where multimodal AI models are used to generate realistic parts of, or even entire film or video sequences.
And let’s be honest there, the writing has been on the wall for something like this for a while now. I’m sure you’ve seen various deep fake videos as well as all of the advancements we’re seeing everyday in 3D and animation using AI.
But besides that, what are the characteristics of silicon hollywood that I can foresee?
Well for one thing, I can see it empowering the individual. If you can just talk to a computer, give it text instructions, and have it create your exact vision about what should be happening entirely in a film scene, suddenly the idea of a one-person film production team may actually be possible.
I’m also excited about film and video, for the first time, being a lot less driven by production costs. Special effects or stories which may have costed hundreds of millions of dollars to create may now be achievable with shared access to the same multimodal AI models. Keep in mind, in the past, cost used to drive everything in Hollywood, from costume design, story telling, location, acting talent pool, and more. What stories were ruined or never saw the light of day simply due to financial constraints?
In Silicon Hollywood, you will only be limited by your imagination. If theoretically, anything you can describe with words can be generated by AI, suddenly the whole realm of storytelling opens up to you right at fingertips. What will you come up with?
I also think we’ll have more time to spend on advanced creative ideas like worldbuilding. Imagine being able to choose from multiple environments created by AI instantly and dicating all the rules behind the universe, characters you want created for your film. You’ll be able to spend more time on concept art. There will be more room for exploring abstract concepts never before seen in film, or new kinds of story arcs which were simply too challenging and costly to explore in our live-action film based past.
I would also add, there is a common saying in Hollywood, that a movie is reinvented three times. First in the writer’s room for the screenplay, then during principle photography, then again finally in the editing room floor. I no longer agree with this sentiment. With multimodal AI, you may finally be able to craft your exact vision in a single integrated stage. If everything is digital, you can retain more control over your work and build exactly what’s it in your head. All three traditional stages - writing, filming, editing could be collapsed into one where you’re simply talking to an AI model and having it generate what you want instantly.
I could go on for days about what else might be possible, but the other things I can imagine is just even fewer traditional hollywood gatekeepers, a lot more experimentation and collaboration amongst film makers, and also a lot less pompousness and exclusivity compared to the traditional Hollywood existence.
Now, let’s get back to today. I am really scared I have been overselling the Silicon Hollywood dream here. Although I think these possibilities will happen soon, I could be super wrong. In the meantime, you will likely deal with technology far more inferior than what I’m describing. You may still have to wait for images to load, only use AI partially, or go through many versions with an AI model before it comes up with something reasonable for your video. In a future video for this series, I hope to outline some Silicon Hollywood benchmarks, technical goals, and industry milestones that I think could be helpful and significant before we consider it as truly something which has, “arrived”.
At the same time, there’s a whole separate AI training data copyright issue I haven’t even touched here … I’m really sorry about that. Is it legally acceptable to train AI on every hollywood movie ever made and use that data to generate entirely new ones? I would say yes, but I’m not a judge or expert copyright lawyer, and something tells me this will have to go through the legal system on its own first before we can have a clear answer.
Finally, don’t get me wrong, I have friends who work in film. Making a living is already hard for most artists and crew in Hollywood. What do these technological shifts mean for the caterer on site? Producer? Stunt coordinator? Struggling actor? Camera operator? It’s unfortunate, I can’t even comment on what this change can mean to their livelihoods, their families, the unions they are a part of, and the clientele they have already built as, perhaps, freelancers over the years.
But I will say this, I often love to meet people who are into different creative endeavours and I always learn that they have always been highly creative even when they were kids and have stories and ideas of their own they want to see come to fruition. Many of them fell into the Hollywood industry with the hope of one day crafting a great film of their own someday. Through multimodal AI, everyone in the film industry may finally be given a chance to craft great multimillion dollar level stories of their own without the need for approval from the hollywood elite or help from anyone in other trades. While hollywood may currently be providing somewhat of a living for many of these individuals, I would pose the question do the creatives actually find it fulfilling? Do they feel accomplished and self actualized by the current status quo in their industry? What great films would they be making if they didn’t need the backing of the industry itself?
What will they create and how will they profit as brilliant creatives once they are liberated, in a way, through multimodal AI?
I apologize if I do not sound compassionate, it’s really hard for me to address this issue of employment, but I still think these are important questions to ask and I sincerely hope for the best for other creatives just like me.
Silicon Hollywood is a new way of approaching the art of filmmaking. Powered by multimodal AI models, it may completely change Hollywood forever.