Open AI sora revolutionizes videography
Updated: Apr 17
By Joshua Yan Apr. 3, 2024
From disco-dancing kangaroos to petri dish pandas, OpenAI’s Sora has recently emerged as a massive pioneering force pushing AI development, being capable of generating videos with lifelike accuracy and sparking both new excitement and concerns with its wide range of possibilities.
OpenAI announced Sora on Feb. 15 alongside almost 50 videos generated from a variety of text prompts about wildlife, city scenes and more. However, OpenAI is currently making Sora available to industry insiders to identify and remediate security weaknesses. A number of visual artists, designers and filmmakers also are giving input to ensure that Sora can be improved to better serve artistic and creative fields. Although the company has not determined a set release date for Sora, some speculate that it will be available to the public some time later this year.
“Though there are a lot of possibilities for possible misuse, Sora could be a mostly beneficial technology depending on how it is used. As long as it is just used to make cool things and not to steal other people’s art, I do not see it as anything particularly scary,” Junior Alexia Lee said.
Despite its potential, the software does have some considerable limitations. According to OpenAI’s website, the model currently struggles with spatial directions, such as specific camera movements or prompts involving rights and lefts. Sora is also not fully capable of rendering certain details for videos with more complex physics, particularly with cause and effect instances, like the missing chunk after a person bites a cookie.
Many of these faults are caused by an underlying problem with generative AI: it does not produce original media. Instead, Sora draws from databases to replicate patterns and anticipate what will happen in the succeeding or preceding frames to create imagery. Because of this, however, Sora is able to extend videos forward and backward in time; this allows the software to create perfectly looping videos and create seamless transitions between two distinct videos.
Researchers at OpenAI believe that being able to simulate the real world with generative AI as Sora does is a monumental step towards developing artificial general intelligence, which seeks to simulate human cognitive abilities such as learning and problem-solving. That kind of technology still remains a dream of science fiction, but in the meantime Sora will serve as a powerful tool in videography and cinematography. Public access to the technology will democratize these fields, but will also potentially hurt existing industry professionals in the process.
Indeed, experts agree that AI will have a massive impact on a variety of fields. A Goldman Sachs report stated that along with its productivity benefits, AI has the potential to force workers out of upwards of 300 million full-time jobs by the end of the decade. Additionally, they believe that the majority of jobs, like those dealing with reception and retail, in the Western world will be assisted or automated with AI to a degree, and that a quarter will be entirely automated.
”AI-generated videos find themselves in a similar position to AI art. Sora and other such video- generation tech like deepfakes need extremely tight regulations, higher than those on AI art. AI-generated videos have an exponentially higher potential in spreading misinformation, meaning users need to be held accountable when utilizing them to prevent the escalation of conflict,” said Sophomore Harry Yin said.
As AI tools such as Sora become increasingly powerful and accessible, they will continue to have massive implications on almost all facets of society, serving as both a convenient utility and gamechanger in artistic, industry and technology-focused fields.
Comments