At least two Chinese developers involved in building Sora, the newly released text-to-video generator from ChatGPT creator OpenAI, have received acclaim on the mainland for their efforts, showing how the advanced generative artificial intelligence (AI) technology has ignited interest across the country where the Microsoft-backed start-up’s services are not officially available.

Unveiled by OpenAI on February 16, Sora can generate videos up to a minute long, while maintaining visual quality and adherence to a user’s prompt. It is able to generate complex scenes with multiple characters, specific types of motion, and accurate details of the subject and background based on what a user has asked for.

Two of the 13 members of the Sora development team, Jing Li and Ricky Wang Yu, were recently singled out on the mainland for their efforts.

Jing joined San Francisco-based Open AI in October 2022, while Wang has only been with the company for about two months, according to their respective LinkedIn profiles.

A video created by Sora, Open AI’s text-to-video generator, plays on a monitor in Washington, DC on February 16, 2024. The Microsoft-backed company said the new platform was currently being tested, but released a few videos of what it was now capable of creating. Photo: Agence France-Presse
Jing was praised by alma mater Wuhan No 2 High School in central Hubei province for “shining on the international stage”, according to a post on Monday in the school’s official WeChat account.

Wang graduated in 2013 from NSFZ, the high school affiliated with Nanjing Normal University in eastern Jiangsu province, according to separate reports last week by Modern Express and Yangtze Evening Post, newspapers which are both backed by the provincial government.

“The innovation from Wang and his team has become a hot topic that continues to attract the attention of teenagers, including NSFZ students”, the Modern Express article said.

OpenAI’s Sora pours ‘cold water’ on China’s AI dreams

China-educated talent is considered as one of the major resources to advance next-generation technologies in the United States, according to MacroPolo, an in-house think tank of the Paulson Institute in Chicago. It found that 27 per cent of AI researchers at American institutions are from China, making up the second-largest geographical group after the US, with 31 per cent.

Jing – one of the principal developers behind Dall-E 3, OpenAI’s most powerful image-generation model – was lauded by his alma mater for being included in Forbes magazine’s 2019 “30 Under 30” list in China and winning the gold medal in the 2010 International Physics Olympiad, according to the Wuhan No 2 High School’s WeChat post.

Before he joined OpenAI, Jing worked for more than two years at Facebook owner Meta Platforms, according to his LinkedIn profile. He co-founded optical computing start-up Lightelligence a year after obtaining his Doctorate of Philosophy in physics from the Massachusetts Institute of Technology in 2019. He graduated from China’s Peking University in 2014 with a bachelor’s degree in physics.
Wang, meanwhile, started his professional career at Meta and its subsidiary Instagram, according to his LinkedIn profile. Wang received his undergraduate education at University of California, Berkeley, right after completing high school.

Chinese entrepreneurs express awe and fear of OpenAI’s Sora video tool

Another Chinese developer could be involved in Sora’s development. But there is little information about Guo Yufei, who is also in the primary product team of Dall-E 3.

Xie Saining, an assistant professor of computer science at Courant Institute of Mathematical Sciences at New York University, has denied his involvement in Sora’s development.

In a widely reported social media post, Xie raised the question of whether China is ready for Sora. He said the country should first make sure that the technology “won’t be abused to serve as a profiteering and manipulation tool by some people or groups”.

But the high quality of videos created by Sora, which is not yet available to the public, astounded many observers when the AI tool initially displayed its capabilities online on February 16.

On its website, OpenAI said the current version has weaknesses. “It may struggle with accurately simulating the physics of a complex scene, and may not understand specific instances of cause and effect,” the company said. “For example, a person might take a bite out of a cookie, but afterward, the cookie may not have a bite mark.”

Source link


Please enter your comment!
Please enter your name here