play videoplay video

OpenAI, the company behind the GPT Chat program and DAL Image Generator, has announced a new tool called “Sora” that is used to create realistic videos of up to one minute in length simply by entering a text request, which is a major innovation in The field of artificial intelligence threatens to change several sectors, according to Agence France-Presse.

The startup company, allied with Microsoft, explained that this new platform, which is based on previous research conducted on the DAL and GPT programs, is still under testing, but it nonetheless showed some video clips and explained how to develop it.

OpenAI explained on its website that the program can create videos of up to one minute in length “with high image quality and following user commands.” She added that Sora can create "complex scenes that include multiple characters, specific types of movements, and precise details." Sora also allows you to create a video from a still image, or expand existing video clips, according to OpenAI.

OpenAI President Sam Altman indicated on the X social network that the company will make the new tool available to a “limited number” of content creators, within the framework of a pilot phase. He also invited users to submit suggestions related to creating video clips, and moments later, he displayed the most prominent of them on the platform.

"Sora" clips

In one of these clips, two dogs are shown fighting in the snow in a mountainous area, and another video clip shows an imaginary animal flying; He's half duck, half dragon, in a gorgeous sunset scene, with a hamster wearing sports gear on his back.

The Sora tool forms the basis for “programs capable of understanding and simulating the real world,” according to the startup, which hopes that the tool will be “an important step in achieving general artificial intelligence,” which is an independent system whose capabilities exceed what humans can do in most tasks.

OpenAI warned that there are “flaws” in the “current model” of the platform, including confusion between left and right or the inability to maintain visual continuity throughout the video.

The company explained that a person in a video clip “may eat part of a biscuit, but the effects of the bite may not appear on the biscuit,” in a second scene of the video clip.

The company confirmed that the issue of security will be fundamental with regard to this tool, noting that it will organize simulations to better define the platform’s boundaries. “We will engage policymakers, educators and artists around the world to understand their concerns and identify positive use cases for this new technology,” she added.

The Meta, Google, and Runway AI groups have previously presented examples of similar text-to-video applications they are developing.

Sora is worrying

The announcement of the artificial intelligence tool Sora raised questions and concerns in the technical innovation and media sectors, after the startup contributed to the spread of artificial intelligence through its “GPT Chat” program.

Among the shots revealed by Open AI is a scene of a fictional creature next to a candle, while an image very close to reality shows a man walking in a wide area. These two examples demonstrate the ability of the Sora program to bring about a major change in the field of video creation.

In an interview with Agence France-Presse, Thomas Bellinger, one of the founders of Bookpack Productions, said: “We have followed the development of the image generation sector, which has created a great internal controversy, and sometimes immediate reactions on the part of the innovators,” adding that “there are those who feel that it is a phrase "It's a storm that can't be stopped and is moving so fast that others don't see it that way."

Cutback Productions specializes in the large-scale use of images and “motion design” for concerts or exhibitions using augmented and virtual reality technology.

“No one has yet tested the new OpenAI product,” Bellinger says, “but what is certain is that someone was expecting this technological tool within a few weeks,” adding, “In the future, we will find different ways to innovate.”

video games

 As for the video games sector, which may witness a radical change as a result of this technological advancement, it shows a division at the current stage. The French company Ubisoft welcomed what it described as “progress,” and a company spokesman told Agence France-Presse: “We have been exploring these possibilities for a long time, and as creators of worlds and stories, we see that there are a large number of future opportunities open to players and our teams to express their imagination and creativity honestly.” Larger".

“I find that using this software at the moment is rather precise,” said Alain Puget, director of the Nantes-based Alchemy studio. “I do not intend to replace my fellow artists with these tools,” stressing that “artificial intelligence is limited to reproducing things that I have done.” Humans".

But he stressed that this “visually impressive” tool could be used by small innovation studios to produce professional-looking images.

While video footage is only a limited part of a video game scenario development process, Puget predicts that "tools like Sora or other generative AI that produce videos will find their way in the long term, replacing our methods of innovation."

The media

Basil Simon - a former journalist and currently a researcher at Stanford University in the United States - points to “a jump that occurred over the past year and was terrifying,” and says that he fears how this tool will be used during electoral periods, and fears that the public will find itself in a position of being “unable to know what to believe.” ".

Julien Bain, presenter of the program "Free or Vu", which is concerned with checking information on the France Info channel, expresses his concern and says: "Until today, it was very easy to detect fake images by noticing that faces are repeated in the background, for example, but it seems that what this new program provides... "It's on another level. We don't have a magic solution."

“Videos can be flagged as AI-generated, and OpenAI can respect that,” he adds. “But what about Chinese or Russian competitors in the future?”

Advertising agencies

As for the “Fred & Fred” agency, which has previously cooperated with brands including “Longchamp” and “Budweiser,” and with a studio dedicated to artificial intelligence that opened in early January, it is expected that “80% of the brands’ content will be created based on artificial intelligence tools.” “This will refocus on creative genius, and production will cease to be a subject.”

Although Stephanie LaPorte, director and founder of the advertising and influence agency OTA, believes that the tool is likely to “strongly change the sector,” she expects “a decline in terms of production,” as companies will resort to these new tools when their budgets are medium or low.

She believes that the luxury products sector may constitute an exception because it is “very sensitive to authenticity, and brands in this field may use artificial intelligence to a limited extent.”

Source: French