Generative AI is already disrupting media and entertainment

Unlike previous waves of artificial intelligence (AI) innovation, generative AI tools seem uniquely capable of being creative.¹ In a short time, generative AI has empowered legions of creators, disrupted industry workflows, and sharply amplified challenges to intellectual property, trust, and ownership. Ultimately, these tools could shape a fundamentally new world of media and entertainment.²

Generative AI is being fueled by investors shifting capital into the tools and companies bringing it to life (figure). In 2022, when general-purpose large language models (LLMs) and sophisticated image generators were released to the public, their ability to create and manipulate content was unleashed. The internet has since been awash in robot philosophy,³ LLM rap battles,⁴ images of clergy in high-fashion puffer jackets,⁵ and “new” songs “sung” by popular artists without their participation or consent.⁶

Actors and their digital twins

Many industries may likely be impacted by generative AI, but it’s already reshaping content creation for media and entertainment.⁷ In film and TV, actors are being paired with their digital twins. By training a generative AI model on an actor’s archive of film, it can “learn” to simulate the actor’s voice, inflections, gestures, gait, and every flinch and tic that makes them unique and recognizable. In one recent example, this technique was used to de-age an actor, not only by modeling their de-aged appearance, but also how their younger self would move and act.⁸ By leveraging real-time virtual production tools, the cast and crew can see the de-aged version as they shoot.

This could change the value of an actor, as the technology extends beyond their physical self to include their digital twin. It’s already possible to “instruct” a digital twin to move and act as needed, nearing photo- and audio-realistic simulation. What, then, is the role of the actor? When is their physical form needed versus using their digital twin? Could some celebrities become “immortal”? Could entirely new synthetic celebrities emerge? The implications for rights and reuse are raising questions and speculation.⁹

In music, such implications are already playing out. Voice cloning has accelerated, enabling the ability to generate new songs and collaborations featuring world-famous artists, or to “rerecord” classic songs with new lyrics sung by voice clones of their singers.¹⁰ Often, these creations have been made without any consent or legal rights to reuse. Similar models are being trained on music in specific styles and from specific artists. As with celebrities, the abilities of musicians can be expanded by generative AI just as their music could become easier to reproduce.

More musicians are leveraging generative audio to not only enhance workflows—and even business plans—but to also discover novel musical elements, like new sounds and rhythms.¹¹ US copyright law is gearing up for a battle to determine not only how training sets may need to protect rights, but also how “human” music must be to be copyrightable.¹² In effect, the history of human creativity—or at least all of it that has been digitized—is now a training set for any number of generative AI models.

Generative game design and conversations with game characters

In video games, generative AI is empowering game development by accelerating ideation and enabling greater realism of worlds and characters.¹³ Visual models are being used to create new characters, set pieces, and visual styles, and even turn 2D renderings into 3D models. For game designers, modeling a small section of the game world by hand can be enough of an input for a game engine to then generate an entire game world.¹⁴ These same game engines now often power the virtual production tools that enable cinematic universes, crossovers between film and gaming, and celebrity digital twins.

Game play is also becoming yet more realistic with generative AI. The behaviors of nonplayer characters (NPCs) have relied on AI for some time, but they could only ever say the few prerecorded lines they were given. Now, leveraging LLMs and voice generation, a recent demo showed natural language interactions with game characters.¹⁵ This enables players to converse with NPCs that can talk freely, generating dialog on the fly that is more casual and creative but consistent with the game narrative.

These tools are evolving quickly and could soon deliver more on-demand experiences, while reshaping the toolchains that enable them. Just ask for a style of music, a flavor of art, or a particular TV or gaming experience and it could be generated on the fly!

Adopt the tools and address the challenges

Successfully navigating this shift may require creators and companies to learn how to use the tools to their advantage—and to face the broader impacts and implications of creative systems that can generate outputs that may have been trained on someone else’s licensed content, that can mimic or falsify voices and faces, and that could present “false truths” that are indiscernible by humans.¹⁶ This tension between use and misuse may likely be challenging to navigate and could soon require more technical, legal, and regulatory guardrails.

Content creators should be willing to embrace these tools and understand how to use them more effectively to augment and amplify their work. Celebrities, rights owners, and studios should push for capabilities and legislation that can establish greater trust, stronger rights management, and legal protections that are crafted specifically for an emerging world of increasingly powerful generative AI. Features and laws will be needed to determine when generative models have used unlicensed training sets; violated name, image, and likeness rights; are fraudulent; or are fully synthetic creations with no discernible human elements. If these practical challenges are addressed, media and entertainment will likely be more empowered to embrace generative AI as a revolutionary way to enhance and amplify human creativity and storytelling.