Sora, OpenAI’s generative video tool that produces studio-quality videos from text, created waves when it was shown last month, and it appears that it will be formally launched sometime this year.
OpenAI CTO Mira Murati told The Wall Street Journal that Sora will be available to the public before the end of this year, and maybe within the next few months. So yet, we’ve only seen pre-built films from OpenAI.
Before it is published, OpenAI is working to make Sora more efficient and affordable. The idea is for it to output video at the same rate that DALL-E produces photos while being reasonably priced.
This means that Sora might be packaged with ChatGPT Plus and made accessible as an API for third-party developers to incorporate into other products, possibly including future versions of video editing software such as Adobe Premiere Pro or Apple Final Cut Pro.
Murati stated that they are also “red teaming” the AI video generator to guarantee that it does not generate inappropriate content and to reduce the possibility of misrepresentation.
People struggle to spot AI video
The announcement of Sora’s potential future release coincided with a recent poll commissioned by Variety, which indicated that most U.S. people couldn’t identify whether a video was Sora or genuine.
HarrisX carried out the poll, which featured OpenAI demonstration movies of waves against steep rocks and the California Gold Rush.
We will not release anything we don’t feel confident on when it comes to how it might affect global elections or other issues.
Miri Murati, OpenAI CTO
Many responses even claimed that AI was used to make human-created movies, implying that OpenAI and other video platforms should focus more on identification.
Murati told The Wall Street Journal that this is an issue that keeps her awake at night, and that they may consider postponing its introduction if safety precautions cannot be implemented before the November presidential election.
“Dealing with issues of missinformation and harmful bias” is critical, Murati stated. According to the company, “we will not release anything we don’t feel confident on when it comes to how it might affect global elections or other issues.”
What will Sora be able to do?
Murati predicts that when Sora is published, it will cost around the same as DALL-E and take around the same amount of time. To get there, they must first improve efficiency and speed, as it now takes “a few minutes” to create a simple 20-second movie.
Murati adds they’re also working on enhancing functionality because Sora still has unrealistic flaws, such as finger movements.
“ChatGPT and DALL-E are optimized for the public to be using them where Sora is a research output so it is much more expensive,” Murati said in an interview with the WSJ. “We don’t know what it will look like when we distribute it to the public, but we expect to acquire it at a comparable cost as DALL-E.”
“We are trying to figure out how to use this technology as a tool people can edit and create with,” he said. “Eventually there will be more steerability, control and accuracy in reflecting the intent of what you want.”
What limitations will Sora have on release?
OpenAI is now engaged in a process known as “red teaming” with Sora, in which they test its limitations and attempt to get it to do things that aren’t ideal, such as making video that may be unlawful, impractical, or potentially defamatory since it portrays public individuals.
The end result will be a model with comparable guardrails to OpenAI’s image model DALL-E. For example, it will not allow you to create videos of public personalities.
Murati says they’re also talking with artists and policymakers from a variety of professions to assess how much flexibility the tool should have over things like nudity and violence.
“I see it as a tool for extending creativity,” she went on to say. “We want people in the film industry and creators everywhere to be a part of informing how we develop it further and how we deploy it.”
“AI tools will extend our creativity and knowledge, collective imagination and ability to do anything but it is going to be extremely hard along the way to figure out the path to bring AI tools into our everyday reality but I think it is worth trying.”
