OpenAI Launches Text-to-Video Tool
OpenAI, the company behind the artificial intelligence (AI) model ChatGPT, Thursday unveiled Sora, which — named after the Japanese work for 'Sky' — can generate videos up to one-minute-long from text prompts....
Facts
- OpenAI, the company behind the artificial intelligence (AI) model ChatGPT, Thursday unveiled Sora, which — named after the Japanese work for 'Sky' — can generate videos up to one-minute-long from text prompts.1
- During a demonstration, OpenAI released minute-long Hollywood-quality videos, including one showing a wooly mammoth running through the snow and one with a scene from the streets of Tokyo.2
- The Tokyo video was generated from a simple text prompt requesting a 'bustling city street,' 'people enjoying the beautiful snowy weather,' and 'gorgeous sakura petals … flying through the wind along with snowflakes.'3
- As part of Sora's release, OpenAI CEO Sam Altman asked followers to submit prompt ideas on X, formerly Twitter. This led to several creative videos, including 'Two golden retrievers podcasting on top of a mountain.'4
- In an effort to limit bad actors from using Sora to make hateful content, misinformation, or fake videos known as 'deepfakes,' OpenAI had experts test Sora to find ways it could possibly be misused.5
- This comes as Google is testing its own text-to-video tool called Lumiere, Meta created one called Emu, and Runway AI is working to use its products to help filmmakers.6
Sources: 1Guardian, 2New York Times, 3New York Post, 4NBC, 5wsj.com and 6Washington Post.
Narratives
- Narrative A, as provided by Gizmodo. OpenAI is again ahead of the curve. The quality of these videos is astounding, especially since the same footage would've taken hours for a human film crew or animator to produce. OpenAI is also being careful because of the political issues these videos could cause. That's why it won't be released to the public until anti-deception tools have been finalized.
- Narrative B, as provided by wsj.com. OpenAI says it's developing anti-deepfake safeguards, but how much can the company be trusted considering the deepfakes that have been created using other technology? Social media companies keep saying they'll create safeguards, but so far nothing has come of those promises.
- Cynical narrative, as provided by The Maneater. OpenAI claims it wants to make the world better, but if it did it wouldn't be stealing human-made art to train its product, or outsourcing work to underpaid laborers overseas. OpenAI might be creating cool technology, but its real goal seems to be to become just another lucratively compensated Silicon Valley tech company.