In June 2024, the world premiere of The Last Screenwriter, a movie with an AI-generated script, was canceled in London due to widespread protests. However, it was far from the end of artificial intelligence in the film industry. Three months later, the AI video startup Runway signed a landmark agreement with Lionsgate, the studio behind the Twilight and John Wick movies. The company will use Lionsgateâs content to train an AI model, and then Lionsgateâs filmmakers and creatives will use the model in upcoming productions. In the same month, 40 years after The Terminator warned of the dangers of ârogueâ AI, its director James Cameron joined the board of artificial intelligence company StabilityAI.
AI presents complex challenges for the movie industry, raising practical and ethical concerns, from its impact on human workers and the environment to intellectual property protection and audience experience. These concerns must be addressed, but itâs unrealistic to expect digital technologies to transform every aspect of life while leaving filmmaking and entertainment untouched. It’s essential to engage in rational discussion and avoid fear-driven narratives.
Here, I explore the conditions under which AI can be effectively integrated into the filmmaking industry. I will apply the same framework used in research on digital transformation KPIs, analyzing the impact on four categories of business objectives: customer engagement, operational efficiency, new sources of value creation, and workforce engagement.
In a rapidly evolving landscape characterized by fragmented and uneven data, I will draw on a diverse yet limited range of examples such as movies, TV series, streaming platforms, and broader entertainment â despite their inherent lack of consistency or uniformity. I welcome any additional insights, perspectives, or feedback: through shared understanding, we can better navigate these complexities.
Can AI engage viewers like traditional films or even better? Traditional filmmaking is still superior in terms of storytelling and technical movie quality. However, AI is progressing fast and could pave the way for unprecedented types of consumption thanks to versioning. Letâs get into the details.
Who can write the plot of a movie? Is it exclusively a human task, or can AI do it? The fight against AI in the entertainment sector has much to do with creativity. But how can we define it? How can we define original content?
For instance, if a video show is adapted from a pre-existing book, can it be considered original? The Netflix catalog includes over 200 titles based on books. Of more than 3,000 series available globally in the second half of 2023, 26% were season two and beyond, i.e., not brand new.
The proliferation of spinoffs, prequels, and sequels indicates that the industry often bets on âsafeâ success, maximizing the return on previous investments and reducing risk, which is high. Itâs estimated that scripts have a 98% chance of commercial failure. In other words, conventional filmmaking relies on human creativity, but business limits its expression.
So, can artificial intelligence do better? Initial evidence is that human creativity cannot be replaced. In a recent study by the University of Exeter, 293 participants (âwritersâ) were asked to write an eight-sentence story appropriate for teenagers and young adults. A third of the participants received no assistance, but the others used generative AI to get ideas. Finally, 600 other participants read and evaluated the resulting stories based on novelty, usefulness, and emotional characteristics. The research conclusions are clear: AI helps writers who lack inherent flair but results in similar stories and fewer original ideas.
Field experiences in the entertainment industry confirm this. Some comedians use AI to write jokes, but although the tool is useful for brainstorming and testing, it cannot yet replicate or replace human creativity.
In a nutshell, blockbusters need human authors, but AI can be integrated for limited support. KPIs like Net Promotor Score (NPS) and the share of AI-enabled films could measure success.
âIf AI-generated films also mean AI-generated actors, will the audience reject them as fake? Will people expect a real human to interpret life, dreams, and feelings on film, or will they decide that a movie is fiction, and a digital actor is fine?â
In the short term, AI cannot disrupt how people make movies and TV. Video generators are limited and impractical â simply put, prompts require text, and describing motions with words is difficult. However, long-term changes could be beyond imagination, according to professional filmmakers and VFX experts. The visual quality is rapidly improving and, sooner than expected, we might see a new genre between films and video games with more interactive storytelling.
If AI-generated films also mean AI-generated actors, will the audience reject them as fake? Will people expect a real human to interpret life, dreams, and feelings on film, or will they decide that a movie is fiction, and a digital actor is fine? The popularity of cartoons and Japanese anime indicates the latter. Filmmakers could build assorted actor crews perfectly adapted to specific segments based on geography, age, and more.
It will take time before we see hosts of AI-generated actors on screens. However, this might happen and again KPIs like NPS and share of AI-enabled films could indicate acceptance.
Digital technologies can engage viewers via personalized recommendations. Netflix, for example, proposes movies and series based on parameters such as viewing history, ratings, title information (genre, categories, actors, release year, etc.), time and day of watching, devices used, and length of watching. If this works with traditional filmmaking, the potential is even higher with AI-powered movies. Versioning would allow for the almost instantaneous creation of multiple releases based on recommendation algorithms â length adjustment, scene assortment, color shading, and music adaptation would be much easier digitally.
Additional views through recommendations would be a valid KPI of AI integration.
Can AI improve operational efficiency in the movie industry in a sustainable way? While the overall environmental impact calls for further analysis, itâs evident that AI can massively reduce costs and accelerate times. Going forward, AI might be integrated into traditional filmmaking for specific elements, like scenes with many actors and complex sets, travel-intense scripts, bad weather, and animal participation. Letâs have a closer look.
Traditional filmmaking is very expensive. Oppenheimer cost around $100m before promotional spending. Television isnât any cheaper: a single episode of Stranger Things costs an average of $30m and The Witcher $10m. Costs include crew (actors, director, etc.), sets, locations, costumes, transport, insurance, script development, equipment, technology, and visual effects.
AI could radically change these economics. For a fully digital movie, AI could lower expenses for travel and accommodation, minimize transport of equipment and costumes, make some forms of insurance redundant, and even save part of the crewâs salaries. The limitations of AI video generators must still be overcome, but a simple comparison of a 60-minute film leaves no room for doubt: with a subscription plan at $229 for 20 minutes and additional minutes at $15 each, one hour would amount to only $829.
Average budget saving would be a reliable KPI to measure the benefits of AI integration.
Seasonality, festivals, and other events determine success or fiasco. Time is precious and delays are dangerous. Besides the actual filming, movies and series require pre-production and post-production work, and the process can be long and troublesome.
Pre-production includes planning, hiring the crew, casting, finding locations, and building sets â pre-production of a Netflix series can take up to six months. Regarding actual filming, many factors influence its length. For example, a long script can involve several scenes with dedicated set-ups and shoot days; night scenes demand more time; bad weather delays exterior filming; scenes with animals are never easy, and actors can make mistakes when enacting scenes. Finally, post-production includes editing, color finishing, sound mixing, music composition, and visual effects. For a movie, it can take up to a year.
With AI, the overall timing is reshaped â no issues with weather or night scenes. Parts of pre-production and post-production are ânatively integrated.â Depending on the AI capabilities and the computational resources available, a one-hour video could take just a few days to generate.
KPIs like speed on time-to-market and error reduction could measure the results of AI integration.
New value might be built around an AI-character ecosystem that could leverage physical-digital integration and brand extension, enabling innovative subscription models.
In traditional filmmaking, crews have a large environmental impact. For example, more than 3,300 people worked on Iron Man 3. The whole circus requires travel and accommodation, heating and cooling, lighting, materials for sets, equipment, transport, and logistics. Fuel, mainly used for vehicles and generators, is the biggest contributor to emissions.
The carbon footprint of a conventional film ranges from almost 400 tonnes of CO2 emissions (the equivalent of a petrol-engine car driving 1.6 million kilometers) to over 3,000 tonnes. A one-hour episode of a TV series has a footprint of 77 tonnes. So, how about the carbon footprint of AI videos? For a 60-minute episode, it could be less than a tonne, based on estimated image number and energy consumption. Movies are normally played at 24 frames per second (fps), which in an hour (3,600 seconds) means 86,400 images. The energy consumption for one image is comparable to charging a smartphone and depends on model efficiency and image size. However, we can assume that an AI-generated image produces up to 0.5 grams of CO2. If we multiply this value by 86,400 and remember that a million grams equals a tonne, the final result is 0.0432 tonnes of CO2.
This value is remarkably lower than the 77 tonnes of traditional filmmaking, but unfortunately, it covers only the footprint for using AI and not for training it. Researchers at the University of Massachusetts Amherst have estimated that the carbon footprint of training a single big language model is around 300 tonnes, the equivalent of 125 round-trip flights between New York and Beijing. On top of energy, we should add water consumption: the University of California, Riverside, has assessed that running some 20 to 50 queries on ChatGPT destroys a half liter of fresh water in the form of steam emissions.
While concluding that AI is also very polluting, itâs worth reporting enhancements like smaller and more efficient models, clean-energy usage, and detail-sharing for developers to build on existing work. Furthermore, data center hardware and cooling systems are improving â companies like Sustainable Metal Cloud (SMC) can reduce energy consumption for air cooling by up to 50%.
To assess AI integration in terms of sustainability, reducing the carbon footprint is a mandatory KPI.
Can filmmakers find new sources of revenue and profit, thanks to AI? Going beyond physical barriers brings unprecedented opportunities but requires creativity and involves risks and, as we enter uncharted territory, we cannot take too many cues from the past. New value might be built around an AI-character ecosystem that could leverage physical-digital integration and brand extension, enabling innovative subscription models. Let’s explore a few options.
Show business resembles an ecosystem built around human actors â they embody peopleâs feelings, ambitions, and lives, choosing productions and leveraging PR to represent characters appreciated by certain types of viewers. Why would this not be possible with digital actors whose features can be tailored to individual preferences without risks of scandals and mistakes? After all, fictional characters have already entered reality via cosplay.
An AI-character ecosystem would find new sources of value in breadth and depth. Character breadth means that the same digital actor could be an avatar or a music star, appear in virtual magazines, be an educator or a training coach, participate in gaming initiatives, or present TV shows. Character depth means that, like a Barbie doll, different versions could be adapted to various segments. Hence the music star could be a rapper or a rocker, appealing to specific ages and performing in any language with perfect pronunciation. Filmmakers would offer a whole ecosystem.
The wider entertainment industry offers several examples that go in the ecosystem direction. Regarding breadth, Meta has paid up to $5m to celebrities like Snoop Dogg to create their AI personas. An animated TV series with kid-friendly versions of DC and Marvel superheroes recall character depth. In 2022, the actor James Earl Jones signed over the rights to Lucasfilm to recreate his Darth Vader voice via AI: so much potential.
When looking for new sources of value, success cannot be taken for granted. The Meta initiative has not gained the expected traction, but the idea should not be dismissed too hastily. Perhaps the error lay in digitizing pre-existing real-world stars like Snoop Dogg, with 88.4 million Instagram followers. However, if objectives are clear and performance is duly tracked, results can be great, as shown by Cadburyâs AI-powered ad with Indian movie superstar Shah Rukh Khan (about +30% sales, over 105,000 user-shopkeepers, and +22% Video View-Through Rate).
The number of AI characters, in breadth and depth, would be an effective KPI to measure results.
âWith AI, pursuing such opportunities might become easier and faster: filmmakers could tap into the existing awareness and loyalty of viewers without the hassle of the production phases, like set-building, equipment, transport, and much more.â
LEGO is an example of integration from the real world to the digital one, as it moved from plastic building blocks to video games, mobile apps, and virtual design. Another example is Monopoly GO!, the digital adaptation of the iconic board game that has achieved $2bn in gross revenues and over eight million daily players.
But itâs also possible to integrate the other way around, i.e., from intangible to physical: we have seen this for many years with Japanese anime and manga brought to life by Bandai (think about all the Gundam robots in toy shops). Similarly, the AI-character ecosystem could be further boosted by a physical-digital integration enabled by VR-AR solutions that integrate virtual actors with viewers and other real people.
As a KPI, combined physical revenues would be a valid support.
For traditional filmmakers, spin-offs, prequels, and sequels are regular sources of additional revenue and profit comparable to brand extensions. With AI, pursuing such opportunities might become easier and faster: filmmakers could tap into the existing awareness and loyalty of viewers without the hassle of the production phases, like set-building, equipment, transport, and much more.
The profit from brand extension could be the right KPI to assess the result of AI integration.
In the beginning, there were movie subscriptions. Then came streaming services. Coming soon, with an AI-based ecosystem, why not subscribe to a favorite digital character? A viewer could select specific genres, including prequels and sequels, and add music concerts to the package. Furthermore, fans could adopt digital actors as avatars to teach them languages or guide them as personal trainers.
The average revenues per AI-character subscription would be a reliable KPI.
Can artificial intelligence support the workforce, improving its productivity and satisfaction? AI cannot replace humans, but it can be an excellent copilot. This is already happening as solutions are commercially available and prove effective. Letâs see how.
The company Jumpcut Media represents a practical example of how AI can help streamline at scale the evaluation of ideas. Over 50,000 new scripts are registered annually with the Writers Guild of America, and producers and agents must assess their potential. How to perform this daunting task? In traditional filmmaking, the first solution is old-fashioned and biased: a script simply has a better shot if it comes from a respected agent or writer. As an alternative, interns read and do the sifting, although they lack the qualifications and expertise.
Artificial intelligence radically changes the approach. In less than two minutes, Jumpcutâs AI solution reads a script and generates two-page reports for producers, breaking down genre, subgenre, logline, synopsis, plot highlights, characters, and similar titles. This doesnât replace human creativity but helps professionals figure out if an idea is interesting and deserves an in-depth study. It saves time and can improve quality.
Additionally, the machine is (at least in theory) not biased: this can concretely contribute to DEI (Diversity, Equity, Inclusion).
Time-saving, percentage of approved scripts, and even DEI rate are KPIs to evaluate results.
The Jumpcut case goes beyond productivity: AI frees people from repetitive work, so humans can finally stop working like machines.
In art and entertainment, a special meaning of workforce satisfaction has to do with the protection of intellectual property. Digital technology can scrape and emulate creative work without permission, which is a serious pain point. In the case of literary classics, the copyright may have expired (for the Berne Convention, the minimum duration of copyright is for the life of the author and 50 years after his or her death).
But things are totally different if, for instance, companies feed recent artwork into AI systems to âtrainâ image-generators. Technology might cause the problem, but it can also offer a solution: an example comes from the University of Chicago and relates to art and generative AI. Researchers have created software that can scramble what AI sees and subtly change images. As a consequence, AI perceives a different art style and training models get confused about image contents. Programs like Glaze and Nightshade are paving the way for effective protection of human work and creativity.
The integration effectiveness can be measured via KPIs like employee NPS and workforce usage rate.
Historically, technological innovations have often been met with skepticism, yet many have ultimately created new industries and jobs.
The table below compares traditional filmmaking to an AI-based model. Itâs a simplification, subject to possible modification in actual cases, but it highlights huge possibilities for integration. While it cannot replace humans, AI can be leveraged as a powerful support â in the short term, increasing efficiency and improving workforce satisfaction, in the relatively near future contributing to viewer engagement and offering new sources of value.
Besides bringing great opportunities, the digital era also raises existential questions. Are digital actors good for our society (because entertainment is about peopleâs dreams), or is there a risk of isolation and alienation? Will AI democratize filmmaking or strengthen the market power of a few players who may be entering from other industries?
Time will tell. However, we cannot expect the competitive landscape to remain as it is now â we have seen that with the music industry. Historically, technological innovations have often been met with skepticism, yet many have ultimately created new industries and jobs. The key to a positive integration of AI lies in establishing clear ethical guidelines, protecting human work, and safeguarding the environment. By doing so, we can ensure that this technology complements human intelligence, enhancing our abilities and freeing us from routine tasks while still valuing and preserving the unique contributions that only humans can make.
If we do the right thing, AI in filmmaking will not be a scary movie.
The quicksilver evolution of AI technologies is opening extraordinary opportunities but leaves human beings exposed to risks. Is it time we developed Artificial Integrity to safeguard fairness and security even as we...
Vital to trade and commerce throughout history, trust is more important than ever in the age of global warming and artificial intelligence...
Here are six ways in which the role of a manager may evolve in an age of advanced AI....
Michael Wade and Tomoko Yokoi explain how IMD ranked the worldâs leading companies on AI adoption and use â and reveal which topped the list...
Explore first person business intelligence from top minds curated for a global executive audience