Kunal Shah, founder and CEO of fintech unicorn Cred, received a reply from OpenAI CEO Sam Altman on Thursday, following the request of the Bengaluru-based entrepreneur, asking for a particular video created using OpenAI’s text-to-video model Sora.
Sora enables users to create realistic videos using a simple text input. The tool is yet to launch in the public domain.
Altman, reached out on X, asking users for input on the kinds of videos generated by Sora they’d like to see. He encouraged his followers to share specific captions for the videos they wished to be created.
Kunal Shah was amongst the numerous people who responded.
He shared his vision of a bicycle race at sea, where various animals would serve as cyclists, captured from a drone camera’s perspective. “A bicycle race on the ocean with different animals as athletes riding the bicycles with drone camera view,” Shah wrote, sharing the video he wanted to see.
Sam Altman responded to Shah’s request, sharing a video generated using Sora, depicting exactly what Shah had written. The captivating 10-second clip shows penguins and dolphins race in the ocean on bicycles. Shah later acknowledged the video with a ‘like’.
https://t.co/qbj02M4ng8 pic.twitter.com/EvngqF2ZIX
— Sam Altman (@sama) February 15, 2024
OpenAI, the company behind creating ChatGPT and image generator DALL-E, stated that they were currently in the testing phase with Sora. They also released some videos demonstrating what the tool was capable of producing, along with the text inputs used to generate each video.
According to a blog post by OpenAI, Sora has the ability to generate videos up to one minute in duration while maintaining the visual quality and following the user’s instructions.
The artificial intelligence (AI) firm stated that it’s engaging in discussions with artists and policymakers before releasing the new tool to the public. “We are working with red teamers, domain experts in areas like misinformation, hateful content and bias, who will be adversarially testing the model,” added the company.
The company is also developing tools to detect deceptive content, including an identification classifier capable of discriminating whether a video was created using Sora.