OpenAI, a leading figure in artificial intelligence, has introduced a new AI model named Sora, which it asserts can produce “realistic” and “imaginative” 60-second videos based on concise text prompts.
According to a blog post released on Wednesday, Sora has the capability to generate videos up to one minute long from textual instructions, including scenes featuring multiple characters, specific movements, and intricate background elements.
The post states, “The model understands not only what the user has asked for in the prompt, but also how those things exist in the physical world.”
OpenAI intends to further train these AI models to aid individuals in solving real-world problems through interaction.
This unveiling marks OpenAI’s latest endeavor following the success of ChatGPT, a viral chatbot, as the company continues to propel the generative AI movement forward.
Senior analyst Reece Hayden from ABI Research notes the novelty of Sora, highlighting its claimed length and accuracy.
Hayden suggests that such AI models could significantly impact digital entertainment markets by providing personalized content across various platforms, particularly within television narratives.
However, OpenAI acknowledges that Sora is a work in progress, identifying “weaknesses” such as spatial details and cause-and-effect relationships within prompts.
For instance, the model may struggle with depicting the aftermath of a character taking a bite out of a cookie without immediately showing a bite mark.
Emphasizing safety, OpenAI plans to collaborate with experts to scrutinize potential risks associated with the new model, including misinformation, hateful content, and bias.
Initially, Sora will be accessible to cybersecurity professors, referred to as “red teamers,” for evaluation of potential harms.
Additionally, access will be granted to visual artists, designers, and filmmakers to gather feedback on potential creative applications.
This announcement coincides with OpenAI’s ongoing enhancements to ChatGPT.
Recently, the company announced a feature allowing users to control ChatGPT’s memory, enabling personalized future conversations or the ability to forget previous discussions.
Source – CNN
