On Feb. 15, OpenAI unveiled Sora, a system that generates videos based on a prompt. OpenAI’s demonstration included short, high-quality videos created in minutes. Video examples included a woman walking through Tokyo and an animated scene of a fluffy monster kneeling next to a melting red candle.
OpenAI named the system Sora after the Japanese word for sky. OpenAI researchers Tim Brooks and Bill Peebles wanted Sora to “evoke the idea of limitless creative potential.” OpenAI, the company behind ChatGPT and DALL-E, is one of many companies competing against Google and Meta to create the most efficient and useful AI products. The demonstration gave a preview of the capabilities of this technology.
Although Sora’s results have been impressive, there are still some flaws, especially with generating videos involving humans and animals in cause and effect relationships. Some people point at these flaws as fundamental limitations of AI technology.
OpenAI has yet to make the technology publicly available as it continues to review the implications of Sora and its dangers. The company has shared Sora with a group of researchers to “red team” the product, meaning they will look for ways it can be misused.
OpenAI has started putting watermarks on videos to identify them as being AI generated, a move to make AI use more regulated. However, these watermarks have been recognized as a weak measure to control disinformation, and OpenAI has stated that they will look for other methods to more strongly curb its misuse.
Tech rivals have taken advantage of OpenAI’s Sora by using the Sora to further develop more advanced videos. For example, on Feb. 27, Alibaba’s EMO, an AI system that adds audio components to videos, used Sora’s video of a lady walking to generate a video of the lady singing Dua Lipa’s songs while walking.
“I think with technology this powerful, it is bound to be used for immoral purposes,” Russell Jin (10), technology enthusiast, said. “With uncertainty, people’s perception of AI will be polarized. Not being able to trust the information on the internet, my skepticism will definitely increase. Information will lose its significance and credibility if it is unable to be distinguished as AI generated.”
OpenAI’s Sora represents a significant advancement in AI technology; however, it is still in the developmental phase. OpenAI’s cautious approach to its release demonstrates the company’s commitment toward ethical AI development.
