I recently researched the oldest film in existence. roundhay garden scene, Four figures, two men and two women, walk with unsteady steps around the garden. It lasts about 2 seconds.
I also recently watched some clips of one of the first fully artificial intelligence-generated videos, created in 2016 by researchers at the Massachusetts Institute of Technology and the University of Maryland. Each is approximately 1 second long. In one, a blurred figure stands on a golf green, bent over to make a putt. Don't mix up these videos or roundhay garden scene For the smooth realism of modern cinema. And just as skeptics often deride AI video as useless, 19th-century critics dismissed early films as “foolish curiosities.”
But a recent agreement between Disney and OpenAI offers a glimpse of a different future. Starting in early 2026, the company's video generator Sora will support Disney, Marvel, Pixar, star wars franchise. Disney+ then streams a selection of user-generated clips.
About supporting science journalism
If you enjoyed this article, please consider supporting our award-winning journalism. Currently subscribing. By subscribing, you help ensure future generations of influential stories about the discoveries and ideas that shape the world today.
Disney will also invest $1 billion in OpenAI and use its tools to build “new experiences for Disney+ subscribers,” according to a joint press release from Disney and OpenAI. “We will thoughtfully and responsibly expand the scope of storytelling through generative AI,” Disney CEO Robert Iger said in announcing the partnership. He also said during a recent earnings call that subscribers will be creating content within Disney+ itself. If you want to see Elsa and Cinderella defeat Maleficent, you can request that scene. However, it may only last 20 seconds.
If this is the beginning of AI TV on Demand, I wonder how long it will take for these clips to reach 20 minutes or an hour, considering the environmental toll and computing costs. Many people believe that it is impossible, but I think very few people who saw this thought so. roundhay garden scene Foresaw big train robbery, 12 minutes of groundbreaking 1903 silent film, much less. Gone with the wind— or streaming.
The challenge with image generation lies in how today's systems work. They are built on diffusion, a technique that starts with “noise” and is gradually refined into an image. Picture a person standing in the fog. The AI essentially removes the fog and inserts new pixels in repeated passes until a consistent shape appears. Each pass to improve the generated image increases cost.
Video is even more challenging. The series of images must be adjusted so that facial features do not change or the coffee mug disappears. Millions of pixels change every second of high-definition video. “We discovered how painful it is to work with video data. These videos have a lot of pixels,” Bill Peebles, an OpenAI researcher who helped develop Sora, said in a keynote at a hackathon hosted by AI community hub AGI House.
To manage pixels, OpenAI's system compresses the video into a simplified version that retains important information. Then treat it like a loaf of bread, slicing it into frames and then dividing it into cubes. This allows the model that powers ChatGPT to align all the cubes with each other so that it associates all the words in the response.
The more frames you add, the more information the model needs to keep visible, so the jump from seconds to minutes can be quite painful. As the video gets longer, inconsistencies accumulate. True “on-demand” AI TV will also require cuts between scenes. If all Disney+ users demanded it with short-term technology, the costs would be huge.
Researchers have been searching for more efficient approaches. First, the model divides the job into stages. “Instead of denoising or generating the entire video at once, we generate it frame by frame,” says Tianwei Ying, a research scientist at AI image editing startup Reve, who co-developed the CausVid video generation software. “At each step, the computation is limited to a smaller portion rather than the whole, allowing for longer processing times.”
Yin believes the system will become more efficient at reaching five minutes of generation by next year, and by integrating various existing AI technologies, it could reach one hour before too long. Others share this optimism. In a recent interview with the BBC, Google CEO Sundar Pichai discussed the possibility of high school students producing feature-length AI films within the next few years. said Cristóbal Valenzuela, CEO of AI video generation company Runway. El Pais “It's not yet possible to make a 60- or 90-minute movie with consistent characters and story, but it's going to happen soon,” he said earlier this month, adding that the company is also looking at watching AI videos generated in real time.
The journey from carefully selected fan clips to feature films goes through some low-key innovations, not to mention negotiations over how to pay the creators who rely on them for their work. And while the financial burden of AI video may seem prohibitive, millions of people around the world are involved in creating and training AI models, and the cost of the technology typically decreases. For example, in 1998, bandwidth was very expensive, costing about $1,200 per megabit per second (Mbps) per month for large networks. However, by 2025, the lowest reported cost was $0.05 per Mbps per month, a decrease of 99.996 percent. This change allows streaming on Disney+ or Netflix.
The cultural path of new media is much more difficult to imagine, and resistance is often fierce. Poet Charles Baudelaire lambasted photography in 1859 as a lazy realism that divorces art from imagination. For centuries, “skeptics and partisans alike compared photography to painting and moving images to theater,” writes contemporary scholar Ruben de Lautour. We seem to be in an even more complex period. What is certain is that technology will evolve as rapidly as ever, allowing millions of creators to test possibilities we cannot yet foresee.
It's time to stand up for science
If you liked this article, please support us. scientific american has served as a champion of science and industry for 180 years, and now may be the most important moment in its two-century history.
I scientific american I've been a subscriber since I was 12 years old, and it's helped shape the way I see the world. siam It always educates me, entertains me, and leaves me in awe of our vast and beautiful universe. I hope that's the case for you too.
If you Subscribe scientific americanhelp us keep our coverage focused on meaningful research and discovery. Having the resources to report on decisions that threaten laboratories across the United States. And at a time when the value of science itself is often not recognized, we support both budding and working scientists.
In return, you get important news. Engaging podcasts, great infographics, Newsletters you can't miss, videos you can't miss, Challenging games, and the best writing and reporting in science. you can too Gift a subscription to someone.
There has never been a more important time for us to stand up and show why science matters. We hope you will support us in that mission.
