Welcome to the /promptcollective!
This week, we're zooming in on a groundbreaking development in AI video technology. You've probably heard the buzz about OpenAI's Sora and Google Alphabet's Gemini 1.5. With Sam Altman of OpenAI casually dropping the news about Sora on February 15th—just ahead of Gemini 1.5—it's clear the race is on. But here at the /promptcollective, we're as always more intrigued by what these advancements mean for the fusion of creativity and AI, than in big tech business beef.
What is Sora?
In essence, Sora transforms text into videos, creating detailed scenes from simple prompts. OpenAI describes Sora as capable of generating videos with intricate scenes and motions just from text. For now, Sora can produce 60-second clips and isn't yet public. The production is quite heavy computational and it takes roughly an hour to generate a clip. And yes, like all new tech, it's got its quirks, like occasionally misshapen hands which was also the case in early Midjourney days.
The Creative Impact
After the initial awe of seeing high-quality AI-generated videos, we started pondering what Sora means for creative industries. Sora could for one revolutionize the ideation phase, how artists sketch out ideas, making it quicker to bring visions to life, whether for personal projects or professional pitches. We foresee a future where creating teaser videos, or 'sizzles,' becomes even more accessible and cost-effective, potentially heralding the rise of specialized 'Sora prompt engineers.'
Rethinking Production
The quality of AI-generated images has soared in just 15 months, and the Sora videos are impressively realistic. Imagine the possibilities if we no longer needed expensive shoots for breathtaking scenes, less say the need to travel to faraway places to hurl Tom Cruise off a cliff on a motorcycle. Movies like the most recent installment in the "Mission Impossible" series could get its adrenaline-pumping shots without the high-stakes stunts and so could all other movies. In a world where such images are available to everyone exactly the presence of big bankable names like the mentioned Mr. Cruise might become even more important to attract the attention of viewers. Impressive SFX/VFX would in this scenario not be a selling point anymore and car chases on mountain roads could become omnipresent:
The Bigger Picture
As AI video creation becomes democratized to a certain extent, one potential development could be a shift towards storytelling and originality. The thinking would be if the bar is raised or if anything can be created instantaneously on one front (the visual side) the bar should be equally raised on the other (storytelling). In that event, we might see filmmakers and visual storytellers explore new rights in an attempt to make their movies or shows stand out.
For the animation industry, AI could streamline production but also disrupt traditional job roles, with people like former Dreamworks CEO Jeffrey Katzenberg predicting significant shifts in employment happening very quickly:
Will it all look the same?
If everyone implements AI-generated video images in their production there is of course the problem that everything might start looking the same. This will be the job of the creatives and producers to control and will still require a lot of skill in our opinion.
The Battle for Attention
In a world where anyone can create stunning video content at a much lower cost than previously, the fight for the viewers' attention will intensify. Ben Thompson, founder of Stratechery, says in this interview in The Town podcast that with potentially more video content available and no scarcity of such high-quality content, the real struggle will be between the big media companies who will fight to become the entry point for consumers and to control their attention on their owned and operated platform. This is a macro perspective on the subject at hand and these companies would still need to make their value proposition to the viewers stand out from the others to get the attention.
Navigating the Legal Landscape
With rapid advancements come legal challenges. The key question is whether legislation can keep pace with the swift evolution of these video generation technologies and how copyright will be dealt with moving forward. For the moment it’s important to note that the output from Sora or other image generators are not copyrightable.
Dive Deeper
That was just a few initial thoughts from our side following the launch of Sora just 14 days ago. There’ll surely be more further down the line. For more insights, please do check out this recent report from a Berlinale and Serien Camp think tank where both Hannes and I participated together with industry professionals from across the EU. Here the implications of AI's for our industry were debated.
Who Are We?
The /promptcollective, by Jes Brandhøj and Hannes Jakobsen, is a hub for those fascinated by the intersection of AI and creativity. Got any ideas or questions? Reach out at join@promptcollective.xyz. And please subscribe.