On the 15th, Open AI, the company behind the conversational AI (artificial intelligence) “Chat GPT,” released a video generated by its in-house developed AI (Sora). AI creates a video when you give text instructions about people, backgrounds, etc. First, we will examine the risk of misuse with experts. The timing of when it will be offered as a service has not been disclosed.
It has been revealed that they are developing an AI model called “Sora” that creates videos. It is said that it can create videos of up to 1 minute in length by giving written instructions. On the 15th, Open AI released several videos created using AI, including “A woman walking in Tokyo at night” and “A mammoth walking in a snowy field.”
In 2023, Open AI began a system to verify the safety of AI by involving experts called the “red team.” Video generation AI will also work with experts to assess the negative impact and risks it will have on society. We will proceed with development in collaboration with video producers and designers.
There are concerns that generated AI could be misused, such as by spreading false information. Based on industry-promoted standards, Open AI has demonstrated that when offering video generation AI as a service, it is possible to differentiate it from a real video by displaying a record that it was made by AI.
Chat GPT was released by OpenAI in November of 2022. He is also working on the image generation AI “DALL-E”. He is also working on development in the field of video generation, making use of the research results he has accumulated in interactive services and image generation.
According to the company, Sora can now be used by a “red team” of experts to identify flaws in AI systems, as well as to collect feedback from visual artists, designers and video producers.
It is possible to generate intricate scenes that combine multiple characters, specific types of movement, and precise subject and background details. In addition to generating videos from text prompts, you can also animate still images.
However, since Sora is an AI model that is still under development, there is a possibility that detailed instructions regarding space may be confusing or that it may be difficult to follow a specific camera trajectory.
Related Post https://www.theatlantic.com/technology/archive/2024/02/openai-sora-generated-video-fake/677493/
Previous Post Microsoft and Google(Gemini) chatbots are fabricating Super Bowl statistics