AI-Made Puppy Videos, Anyone? OpenAI's Sora Converts Text Into Lifelike Videos – News18

Over the past two years, there has been significant AI progress from various fronts, including pioneers like OpenAI and the steadily advancing Google, which is establishing itself as an AI powerhouse. As we step into 2024, tech companies are intensifying their efforts to capitalise on this momentum. Notably, Google has garnered attention by launching Gemini 1.5, while OpenAI, too, has unveiled Sora, its latest text-to-video AI model.

As the name suggests, Sora by OpenAI has the ability to bring text-based prompts to life by crafting lifelike scenes. Picture yourself writing a prompt for an imaginative scenario, such as envisioning a cat dancing atop a spaceship. Sora excels in translating these prompts into detailed and realistic videos, as indicated by the information disclosed by the company.

“We’re teaching AI to understand and simulate the physical world in motion, with the goal of training models that help people solve problems that require real-world interaction,” OpenAI said.

It added that Sora can generate videos up to a minute long and maintain visual quality as per the user’s prompt.

If you visit the landing page for Sora, you can see the myriad results that it has posted. One specific example of a drone shot, wherein waves can be seen crashing along Big Sur’s Garay Point Beach, is as realistic as it gets. I, for one, can’t tell the difference. If you handed me this footage without telling me it was AI-made, I wouldn’t know.

OpenAI claims that Sora can generate complex scenes with multiple characters, specific motions, and accurate details of the subject and background.

The company also claims that the “model understands not only what the user has asked for in the prompt but also how those things exist in the physical world.”

A Scary Future?

As dystopian as it seems, OpenAI says that it is taking big steps to make the rollout safe. The company is working with experts to curb misinformation, hateful content, and bias. Plus, it is also working on detecting misleading content and the ability to tell that a video was generated by Sora. Additionally, the safety techniques used for DALL-E 3 will apply to Sora as well.