On Thursday, OpenAI, the developer of ChatGPT, presented a new text-to-video artificial intelligence model called Sora. Although only a few early video examples were shown, Macquarie analysts described it as a “significant technological advance”.
The presentation included several videos, each about a minute long, “completely produced by Sora without any modification,” according to OpenAI.
With the introduction of Sora, OpenAI has “once again demonstrated that its internal product innovation is driving the industry forward, setting high benchmarks for cutting-edge technology,” Macquarie analysts said in their report.
“We believe the key findings of this publication are the ability to create long-term coherent videos, and the model's ability to accurately reflect how objects interact and move in a real-world environment,” they add.
Researchers have identified three key features in Sora's output.
First, it illustrates the dangers faced by startups that rely on one-size-fits-all models to gain competitive advantage, especially with the rise of well-invested firms.
Second, the speed at which artificial intelligence is transforming industries that rely on creativity, such as video production, is faster than expected.
The third point is that Macquarie believes it will be important to see the resolution of the copyright lawsuit brought by the New York Times. The case may provide insight into how courts interpret copyright laws with respect to AI companies. This result could affect the pace of development of artificial intelligence models like Sora, which require large data sets from various media for training.
This article was prepared and translated with the help of artificial intelligence and reviewed by an editor. For more information, see our terms and conditions.
