Home AI OpenAI Reveals Sora, A Very Realistic Text-To-Video AI Model

OpenAI Reveals Sora, A Very Realistic Text-To-Video AI Model

116
0
Vintage camera with paper note on wooden table.
Vintage camera with paper note on wooden table. Image by jcomp on Freepik

If your five-word prompt is: “I’m going to the moon.” Sora could convert it into somebody going to the moon on the back of a Wandering Albatros.

OpenAI has released teaser videos for ‘Sora,’ its latest text-to-video model. The technology has the potential to turn people into a cinematic reality, allowing them to take an everyday experience like commuting to the gym and generating a video that shows people riding turtles along the ocean, all by inputting a prompt and hitting ‘generate.’

Courtesy OpenAI Although the model isn’t yet available to the public, OpenAI’s announcement has reverberated across the internet in a blend of excitement and alarm, sparking millions of social media opinions on the potential uses and abuses of an AI so creative it can defiantly barf words. Some are jumping at the creative potential of the tool to generate new art in areas as diverse as music, literature, film, and video. Others worry about the potential of AI for erasing jobs and spreading misinformation.

The CEO Sam Altman released a selection of videos Sora had created, including a sea turtle riding a bike, a pasta demonstration, and four dogs doing a podcast from the summit of a mountain. OpenAI confirmed that the model is still just a research project, a spokesperson wrote in an e-mail: “We are not making this model broadly available in our products soon.”

OpenAI is one of the companies leading the accelerating generative AI revolution, and Sora is designed to model multiple characters and motions in more realistic ways; in improving AI’s understanding of the physical world and the application of those physical-world rules to problems, this research has implications in areas as diverse as games and medicine.

But even Sora might struggle to generate faithful renditions of complex scenes, producing nonsensical outputs or other forms of distortion. However, some of the demos released by OpenAI feature realistically looking animations, indistinguishable from human-made footage.

With fears about deepfakes and fake news continuing to escalate, agencies such as the Federal Trade Commission might also begin to move on regulating AI-generated content, too. For its part, OpenAI is not slowing down in its efforts to mitigate safety risks; it explained in a news release that it would be hard at work adding tools to flag Sora-generated videos, ”the FTC is committed to using all of its tools to detect, deter, and halt impersonation fraud.” they wrote.

OpenAI emphasizes the risks but says that real-world feedback is necessary to make AI systems better over time. As the technology continues to evolve, transparency and responsible use will be essential in ensuring that it has net-positive effects on the societies it serves.

Source: Etech,The Economic Times February 16, 2024

Author

LEAVE A REPLY

Please enter your comment!
Please enter your name here