Source: Xinhua News Agency
The OpenAI, an American Artificial Intelligence Corporation, located in San Francisco, recently launched a generating artificial intelligence model Sora, which can create videos through text instructions. This breakthrough has consolidated its leading position in the field of artificial intelligence.But at the same time, it also exacerbated concerns about risks such as "deep forgery".
"A glance" may subvert the industry
This is the first time OpenAI has entered the field of artificial intelligence video generation.According to the company, Sora uses the Transformer architecture to create similar and imaginative scenes according to the text instructions, generate a variety of styles, different formats, and the longest high -definition videos.This model can also generate videos based on static images, or extend or fill in the lack of frames of existing videos.
Some analysts believe that Sora once again highlights the profound impact of artificial intelligence technology progress on real life and traditional industries.The huge development prospects of artificial intelligence in the field of video generation, while opening the door to shaping the new format of the film and television industry, may subvert the current film and television industry.Hollywood encountered the first screenwriter and actor industry in the past 63 years, because some of the job opportunities in the industry may be replaced by artificial intelligence.The birth of Sora made this threat closer and true.On the second day of its launch, the stock price of the main image processing and video production software fell more than 7%.
Although it is not the first company involved in the field of text transfers, the launch of Sora led OpenAI to lead the track and consolidate its leading position in the field of generation of artificial intelligence.Third -party information analysis agency Gaundner Consultant's analyst, Qian Dellasecaran, said: "In this field, no company is more ambitious than OpenAi, and this ambition seems to be increasing."
Openai said that Sora has a deep understanding of language, not only understands the user's text tips, but also the way the existence of things in the physical world."We are teaching the physical world in artificial intelligence and simulation movements. The goal is to train the model to help people solve the problem of interaction with the real world."
"In -depth falsification" is concerned
Sora released can be described as a thousand waves, but it is more like a preview version this time. It is difficult for the public to understand the advantages and disadvantages of the model in depth.OpenAI said that only Sora access rights are mainly provided to some designers and filmmakers such as some designers and filmmakers to obtain feedback on improving the model.Not only did the company not publish basic details such as the data of the training Sora model, but also the date of publishing to the public.
Openai admits that currently Sora generated videos may include unreasonable images, confuse space details such as left and right, and it is difficult to accurately simulate the physical principles and cause and effect instance of complex scenes.For example, a person bite a biscuit, but there was no bite on the biscuits.
However, with the enhancement of computing power and model improvement, people may have achieved a more complete and advanced video in the short term.
Many people in the industry are worried that Sora will help the "DeepFake" technology.Faride, deputy dean of the School of Information College of the University of California, said: "When news, images, audio, videos -can be forgery, then there is nothing in that world." In response to fake concerns, OpenAI said that when launching products to the public, it will ensure that video contains source data, and launch a tool for testing video.OpenAI also promised that before using SORA in the product, security measures will be taken, including experts in the fields of "error information, hatred content and prejudice" to evaluate the harm or risk of confrontation testing by the model;Contents, hatred images, others IP and other text input prompts.
Even so, OpenAI acknowledged that despite conducting extensive research and testing, "we still cannot predict that people will use all the useful ways of our technology and abuse all our techniques."
Supervisory difficulties under the AI rushing
Under the circumstances of subversive innovation in the field of technology, how to achieve technological progress and ensure the balance of social security has attracted more and more attention from all walks of life.
Openai said that it will cooperate with policy makers, educators and artists from all over the world to understand their concerns, determine the active use cases of Sora, and believeA key component of a safe artificial intelligence system.
Industry insiders pointed out that under the circumstances that the current governance framework and management and control measures have not kept up, it is not possible for enterprises to not provide artificial intelligence security and trust required by enterprises.
