How Long Before AI Wins an Academy Award?

Will text-to-video and multimodal models redefined computer generated imagery (CGI)

Welcome to this week’s edition of The Artificially Intelligent Enterprise. Remember to read to the bottom for my weekly bonus.

This is just a reminder that if you haven’t gotten your free tickets yet, go to The AIE Online site and register for free to access some of the brightest minds in Generative AI.

TL;DR - AI News, Tips, and Apps

Feature: How Long Before AI Wins an Academy Award

Tyler Perry spent four years planning an $800 million expansion of his Atlanta studio, which aimed to add 12 soundstages to the 330-acre property. However, he has put his plans on hold due to the rapid advancements in artificial intelligence, including OpenAI’s Sora, a text-to-video model released on Feb. 15. The Sora model impressed many observers with its cinematic outputs. This leads me to wonder how long it will be until we see a feature film winner that was 100% generated by AI.

We are already starting to see GenAI and ML impacting the film industry.

  • AI Predicted Oscar Winners: AI technologies, such as Swarm AI, have shown impressive accuracy in predicting Oscar winners, achieving a 94% accuracy rate in major award categories. This demonstrates the potential and growing capabilities of AI in the film industry.

  • AI's Impact on Film Production: Movie studios are increasingly embracing AI to revolutionize various aspects of film production, distribution, and marketing. AI tools enhance visual effects, streamline editing processes, and even generate scripts, showcasing AI's significant presence and influence in the industry.

  • Growing Influence of AI in Film Market: The influence of AI in the film industry is expected to grow significantly, with the entertainment and media AI market projected to have a compound annual growth rate of over 26.9% from 2022 to 2030. This growth indicates the increasing integration of AI technologies in filmmaking and consumption.

Artificial intelligence (AI) 's rapid advancement in filmmaking, mainly through text-to-video and multimodal models, is not just a trend but a seismic shift that promises to redefine the landscape of computer-generated imagery (CGI). As these technologies evolve, the question isn't if AI will win an Academy Award but when. This blog post explores AI's current filmmaking state, challenges, and potential to revolutionize the industry.

The Current State of AI in Filmmaking

AI's role in filmmaking has steadily grown, from script analysis to post-production and even to the generation of visual effects. Text-to-video models, which generate video content from textual descriptions, are at the forefront of this evolution. These models are trained on large datasets of videos and text descriptions, aiming to create videos spatially and temporally consistent with the input text. 

RunwayML, an advanced video editing software powered by artificial intelligence, played a significant role in producing the critically acclaimed film "Everything Everywhere All at Once," which won seven Oscars.

The film's visual effects artist, Evan Halleck, utilized RunwayML's green screen tool to remove images' backgrounds efficiently, significantly expediting the post-production process. Halleck highlighted the tool's effectiveness, noting that it could cut out elements more accurately than he could manually, providing him with clean mats that could be used for other purposes.

The Rock Universe scene in "Everything Everywhere All At Once" is a pivotal and memorable moment in the film that provides both the characters and the audience with a much-needed respite from the fast-paced and chaotic journey through the multiverse. It was generated with RunwayML. In this scene, Evelyn Quan (played by Michelle Yeoh) and her daughter Joy (played by Stephanie Hsu) are depicted as rocks in a barren, lifeless universe, which starkly contrasts the high-energy and visually complex scenes that precede it.

However, despite these small wins, the technology faces significant challenges, including computational demands, the scarcity of high-quality datasets, and the complexity of video captioning. Despite these hurdles, AI has made notable strides in the film industry. For instance, AI-generated films like The Frost showcase the potential of AI in creating visually compelling narratives entirely from text descriptions. Moreover, generative AI models have been utilized in various aspects of filmmaking, from enhancing visual effects to automating tedious post-production tasks.

Challenges and Limitations

The journey of AI towards Academy recognition is not without its challenges. A significant barrier is the computational complexity of generating long, coherent videos that maintain spatial and temporal consistency. Additionally, the lack of high-quality, diverse datasets limits the ability of AI models to learn complex movement semantics and generate videos that accurately reflect the vast spectrum of human experiences.

Another critical challenge is the ethical and creative implications of AI-generated content. As AI plays a more significant role in content creation, concerns about the originality of AI-generated works and the potential displacement of human creatives have emerged.

The 2023 Writers Guild of America (WGA) strike lasted from May 2 to September 27, 2023, was a significant labor dispute involving 11,500 screenwriters and the Alliance of Motion Picture and Television Producers (AMPTP).

This strike, part of broader Hollywood labor disputes, became the second-longest in WGA history, tied with the 1960 strike and only surpassed by the 1988 strike. The strike was driven by concerns over residuals from streaming media, using artificial intelligence (AI) in scriptwriting, and the overall valuation of writers' work in the evolving digital and streaming landscape.

Screenwriters were particularly concerned about the impact of generative AI on their profession. They feared that AI could be used to replace human writers, undermining their credit, compensation, and separate rights. The WGA's negotiations with the AMPTP focused heavily on establishing clear guidelines for using AI in scriptwriting and other aspects of filmmaking.

The tentative agreement reached between the two parties stipulated that AI-generated material could not be used to undermine a writer's credit or separate rights, and AI could not write or rewrite "literary material."

Moreover, studios and production companies were required to disclose to writers if AI had generated any material given to them. However, the agreement allowed writers to use AI if the company consented, but a company could not require a writer to use AI software.

The strike highlighted the broader concerns within the creative industries about the role of AI. Many experts viewed the screenwriters' deal as a potential model for future labor negotiations in content-creation industries, emphasizing the importance of humans working alongside AI rather than being replaced by it. The agreement acknowledged the rapidly evolving legal landscape around the use of generative AI and left some issues, such as the use of scripts to train AI systems, to be resolved by the legal system.

The strike's resolution brought attention to the need for clear policies and agreements regarding the use of AI in creative processes, ensuring that technology serves as a tool to aid human creativity rather than a means to replace it. The WGA strike and its focus on AI underscored the ongoing dialogue between labor and technology in the digital age, marking a significant moment in the intersection of labor rights, technology, and creative work.

The balance between leveraging AI to enhance creativity and preserving the unique touch of human artists is a delicate one.

The Future of AI in Filmmaking

Despite these challenges, the potential of AI to revolutionize filmmaking is undeniable. AI promises to reduce production costs, enhance creativity, and open up new possibilities for storytelling. AI technologies are expected to become more integrated into the filmmaking process, from pre-production to post-production. One of the most exciting prospects is using AI to create more realistic and complex CGI effects. AI's ability to learn and replicate the nuances of the physical world could lead to CGI that is indistinguishable from reality. Furthermore, AI could democratize filmmaking, allowing independent filmmakers to create high-quality content with limited resources.

Prompt of the Week: Creating Mindmaps

A mind map is a visual diagram representing ideas, tasks, or other concepts linked to and arranged around a central subject. A nonlinear graphical layout enables users to structure information to mirror how the brain processes data. Mind maps are characterized by branches, colors, images, and keywords, all stemming from a central idea to explore a wide range of topics, solutions, tasks, or ideas related to the core concept.

Now, on to the prompt I borrowed and tweaked from Gina Acosta at Horizon AI.

Step 1: Go to ChatGPT and use the following prompt.

You can also make the changes you want to the information or structure. That’s the advantage of having the code.

Create a mind map of [Your Topic]. List topics as central ideas, main branches, and sub-branches.

I got something like an outline here:

Step 2: Transforming Text into Structure

If ChatGPT does not send you your result in a code box, use this prompt:

write the .js needed for MarkMap to create a mindmap

You will get something like this, click on the Copy Code icon in the upper right corner:

Step 3: Go to MarkMap and Visualise with MarkMap

Go to MarkMap and copy and paste your markdown output from chatGPT into the MarkMap editor.

After finishing, download it as an interactive HTML for online use or as an SVG for pictures.

I hope you enjoyed this week’s edition; if you liked it, please forward it to a friend who might benefit.

Best Regards,

Mark R. Hinkle

Mark R. Hinkle
The Artificially Intelligent Enterprise
Follow Me On LinkedIn | Follow Me on Twitter
Follow the AIE on LinkedIn | Follow the AIE on Twitter 

Weekly Bonus: Three Classes to add to your GenAI education

I thought I’d add three easy-to-follow resources for adding to your generative AI toolbox.

Join the conversation

or to participate.