What Sora by OpenAI is and how it works: the new AI capable of creating ultra-realistic videos only with text

OpenAIcreator of ChatGPT and the DALL-E image generator, has introduced a new tool called “Sora”capable of creating, from a simple textual instruction, Cinema-quality video up to one minute in lengtha revolutionary innovation in the field of artificial intelligence.

Built on previous research in the DALL-E and ChatGPT models, this new platformmom it is still in the testing phaseclarified this Thursday to boot The Californian has allied itself with Microsoft, which presented some videos to show its first progress.

What is Sora and how does it work?

According to the first previews shared by Open IA on its website and then massively disseminated on the “X” social network, this technology can create clips up to one minute long “with visual quality and respecting user demand.”

Sora can “spawn complex scenes with several charactersspecific movements and precise details,” the company said.

It also allows create a video from a still image, says the artificial intelligence (AI) giant. Likewise, he has the ability to do so lengthen existing videos.

Sam Altmannnumber one in OpenAI, published on his X account that Sora will be offered “access to a limited number of creators”as part of an experimental phase.

It also invited users to submit proposals to generate videos, after which provided some examples.

Among these videos, you can see images of two dogs climbing in the snow on a mountain. Another clip shows how an imaginary animal begins to fly before a magnificent sunset. A hamster wearing sports clothes rides on his back.

Sora is a base for “programs capable of understanding and simulating the real world”explains the start-up, which hopes that this will be an “important stage in the creation of AGI”, Artificial General Intelligence, a system highly autonomous That would surpass humans in the most economically profitable tasks.

For its part, OpenAI warned that the “current model” of the platform It has some “flaws”such as the confusion between left and right or the inability to maintain visual continuity for the entire duration of the video.

“For example, a person may eat a cookie, but then the cookie may not show a bite,” he exemplified.

The company has specified that security is a key theme in this development and has declared that it will organize simulations with users who will be asked to try to cause errors or create inappropriate content, in order to better define the limits of the platform.

“We will involve policy makers, educators and artists around the world understand their concerns and identify positive use cases of this new technology,” OpenAI said.

Meta, Google and the artificial intelligence of the trackwho work on similar applications, called “text-to-video”, which allow you to convert a written idea into video, also presented examples of their progress.

Hollywood screenwriters at war with artificial intelligence

In May 2023, the mecca of American cinema came to a complete halt. THE screenwriters they started to strikewhich the actors joined in mid-July, to ask for better remuneration and to regulation of the use of artificial intelligence (TO THE).

This unprecedented protest in Hollywood since 1960 ended in September in the case of the screenwriters, who reached a salary agreement and protections against the use of artificial intelligence.

Actors, concerned by fears that movie studios will turn to technology clone your voice and image to reuse them in perpetuity without compensation or consent, they resumed filming in November.

In addition to better wages, the strike succeeded in imposing restrictions on the use of artificial intelligence.

American film and television production was paralyzed for nearly six months by the protest, which cost the country’s economy at least $6 billion.