You may have notice that there is a date right there in the title. With AI moving so fast, this analysis could change in a matter of months, weeks, or even days, so all these ideas could be completely wrong. Either that, or I will be completely correct, and be invited on talk shows and internet podcast to share my insights. I'll settle for somewhere inbetween.
Advancements like OpenAI's announcement of Sora, READ ALL ABOUT IT HERE, in AI technology highlight the growing concerns regarding its impact on employment within the entertainment industry. The ability of AI to produce detailed and complex scenes poses a significant question: what becomes of the traditional roles of cast and crew members, previously essential for filming such content? Will AI replace them all in favor of an experienced prompter?
Courtesty of Sora AI
No Connection Error
While the results are impressive, and will just keep getting better, the one thing AI will struggle with is convincible interactions between characters. It is one thing to have someone walk down a street, or make a cartoon kangaroo dance, but making two characters connect, have chemistry or even just connect with the audience will take a bit more time (or very good prompts) to achieve, if it even can be.
Convincible action, as well as physics, will also take awhile to achieve. Even expensive SFX studios have trouble creating a digital character that has convincing physics (think ragdolls being thrown around). Having two characters interact, shake hands, dance, fight, or even just subtle taps on the shoulder with convincible reactions will be an interesting challenge.
For Every Action . . .
Sora has already admitted that their model has trouble with cause and effect. Their example is a person eating a cookie. The model knows how to make the person, make the cookie, and make the person take a bite of the cookie, but having the cookie then have a bite taken out of it is where the model has trouble.
This would be the same challenge with character interactions. One character would have an action, but creating another characters realistic reaction is where AI may struggle. Take a fight scene, or a romantic kiss. The AI could create the action, but the other characters reaction to it, would be a lot harder to achieve.
Real World Use
While there is much debate about AI taking over the entertainment industry, I think there is still a ways to go to create anything that could compare with a traditional movie or series. I can see these technologies working as more of filler content. Establishing shots or transition scenes, perhaps. Maybe the use of AI to remove or edit parts of a scene (like removing a mustache from Henry Cavil's Superman that looks more convincing).
I can also see AI being highly used, and actually already is, in the video game industry. Not only in the coding and creation of games, but utilizing it into the game itself. Developing a "Name your own adventure" game, where certain parameters are set up by the creators, but where AI steps in to fill in some of the story, dialog, or entire locations and characters. This would add in a whole new level of playability, as each decision, each action, would cause a new path forward for the player. Instead of those paths being dictated by pre-defined parameters, AI can create entirely unique experiences.
In Conclusion . . .
While AI will definitely change the face of image, audio, and video creation, it doesn't mean it will fully take over. It is getting better every day, and before we know it, will be capable of creating high-quality videos that will be indistinguishable to all but those with a very trained eye. This will create a appealing visual, but to achieve the heart, soul and humor people watch movies and TV shows for, it will always need that human touch.
But don't quote me on that.
Comments