OpenAI's New Sora Video Generator Impresses But Raises Familiar AI Concerns
-
Sora is a new video generation model from OpenAI that converts text prompts into realistic videos, but few technical details have been shared about how it works.
-
The quality of Sora's outputs seems heavily reliant on the massive compute resources OpenAI devoted to training it, rather than just innovative techniques.
-
Like other AI products from big tech companies, Sora's announcement came with limited transparency about its training data, potential harms, and more.
-
While Sora's demo videos are impressive, they also contain noticeable flaws in physics, cause-and-effect, and other basics that temper some of the hype.
-
Unable to access Sora themselves, internet commentators are left theorizing about how it works and raising familiar concerns about deepfakes, copyright, and bias.