China News Service, February 19 (China News Financial Reporter Wu Tao and Wu Jiaju) Throw in a piece of text and generate "blockbuster" level video content in minutes? Recently, OpenAI released its first video generation model Sora, which immediately attracted attention.
What does this mean today when short videos are popular? What impact will it bring?
Picture from OpenAI website
AI-generated video effect "explosion"
Several huge woolly mammoths are approaching on the snowy grass, a small furry monster is kneeling next to a melted red candle, a woman is walking on the street... OpenAI said that these almost life-like videos are all directly generated by Sora. Unmodified.
According to OpenAI’s official introduction, Sora can produce videos up to 60 seconds long, including high-definition scenes, complex camera movements, and can support the creation of multiple characters.
OpenAI says the model can understand not only what users ask for in prompts, but also how these things exist in the physical world.
Sora is not currently open to the public for use, but many netizens have praised the "explosion" effect of the released video, and at the same time did not forget to imitate the tone of some "big manufacturers" to ridicule: We have already laid out the plan, (I hope Sora will hurry up open source).
Many netizens are also full of expectations: "I want to see what the real Wanjian Guizong is!" "I want to see what a microscopic ant cave looks like!"
Picture from OpenAI website
Throw out a novel and make a blockbuster?
In addition to generating images that are closer to the real physical world, OpenAI introduced that Sora can ensure that the subject in the video remains consistent, does not change even if it temporarily leaves the field of view, and can generate multi-camera videos at once.
Based on this technology, some netizens have imagined using multiple 1-minute videos to connect to produce a short drama or even a movie, while ensuring that the theme remains unchanged. "Throw a novel out of a blockbuster film" will become a reality.
A research report released by Guosheng Securities on the 17th shows that compared with other previous Wensheng video models, Sora has leapfrogged into a practical productivity tool. The 1-minute length is expected to be widely used in the field of short videos, and the ability to expand videos is also expected to produce long videos.
Zhou Hongyi, founder and chairman of 360, recently publicly stated that this means that the implementation of AGI (artificial general intelligence) will be shortened from 10 years to one year.
Of course, Sora is not without weaknesses. OpenAI says it may have difficulty accurately simulating the physics of complex scenarios and may not understand specific instances of cause-and-effect relationships. For example, a person takes a bite of a cookie, but the cookie may not have a bite mark afterwards.
Additionally, the model may confuse the spatial details of the cues, such as left and right, and may have difficulty accurately describing events over time, such as following a specific camera trajectory.
Picture from OpenAI website
"Overturning" film and television shooting and editing overnight?
But on the other hand, worried voices also have a place online.
Are the video shooters and editors becoming "redundant" when visual quality comparable to that of a "blockbuster" is generated in minutes? Will actors disappear in the future? Is it possible to choose whoever you like to be the leading actor?
Ideas like "opening up your imagination" may not be far from reality.
Some people believe that Sora may "overwhelm" the video editors and photographers overnight, and they will compete to see whose creativity is better in the future. A research report from a securities company stated that this is a milestone event for Vincent Video and may bring about a revolution in the content creation industry.
Shen Yang, a professor at the School of Journalism and Communication at Tsinghua University, said in an interview with a reporter from Sino-Singapore Finance that Sora has a greater impact on film and television-related industries. It can significantly reduce production costs. If it is widely used, the cost can be at least one-tenth of the original cost. , so Sora will have a relatively obvious impact on video, games, education and other industries.
Liu Xingliang, director of the DCCI Internet Research Institute, believes that traditional film and television production processes and business models may also be forced to transform. But he also pointed out that this is an opportunity to dance hand in hand with AI, explore new art forms and expressions, and continue to write the glory of the film and television industry.
Liu Xingliang said that Sora is a super gift package for content creators. It can not only reduce costs and speed up creation, but also enrich the visual experience of the audience. The future picture of AI will be more exciting than we imagined.
Zhou Hongyi believes that AI may not disrupt all industries so quickly, but it can inspire more people’s creativity.
Picture from OpenAI website
How to deal with the risks that AI may bring?
In addition, there are also voices that believe that this new technology may bring about a series of risks and even social problems.
Liu Xingliang pointed out that AI content production blurs the boundaries between reality and virtuality. Issues such as content authenticity, copyright, privacy, data, and security arise one after another. Society needs a complete set of policies, laws and ethical norms to respond to ensure that technological development does not derail and protect everyone's interests.
In this regard, OpenAI stated that several important security measures will be taken before Sora is applied to OpenAI's products. Experts in relevant fields will conduct adversarial testing of the models. The company is also developing tools to help detect misleading content, such as a detection classifier that can tell that a video was generated by Sora.
OpenAI also said it will engage with policymakers, educators and artists around the world to understand their concerns and identify positive application cases for this new technology. Despite extensive research and testing, the Company cannot predict all of the beneficial ways that people may use the Company’s technology, or all of the ways in which people may misuse it.
It is worth noting that OpenAI issued an announcement last year, announcing the global recruitment of "red team" network members, aiming to introduce external forces to unearth the flaws and risks of AI systems in advance.
Prior to this, experts in relevant fields have called on the international community to jointly promote artificial intelligence supervision to ensure its safety.
In November 2023, 28 countries and regions, including China, the United States, the United Kingdom and the European Union, signed the "Bletchley Declaration" at the first Artificial Intelligence Security Summit, agreeing to work together to create an "internationally inclusive" frontier Artificial Intelligence Safety Science Research Network to develop a deeper understanding of AI risks and capabilities that are not yet fully understood. (over)