Text →AI→ Video Opportunities and Challenges Coexist.

Following ChatGPT, recently, OpenAI released the first video generation model "Sora". You only need to input a text instruction to generate a realistic video. However, while the video generation model promotes the development of artificial intelligence technology, it also raises people’s concerns that the deep forgery of video by artificial intelligence may aggravate the spread of false information and false information.

The video you are seeing now is made by the video generation model "Sora". According to official website of OpenAI in the United States, the text-to-video model released this time can generate a video with a duration of up to one minute according to the instructions input by users. For example, according to the text: "A fashionable woman walks on the streets of Tokyo, and the streets are full of neon lights and city signs. She is wearing a long red dress, a black leather jacket and black boots, sunglasses and red lipstick. " You can generate a video like this. The model can also obtain existing static images and generate videos from them, and can also obtain existing videos to expand or fill in missing content.

Before the birth of Sora, Runway, an American artificial intelligence startup, launched a similar technology, but the video produced was short.

Valenzuela, founder of Runway, USA:Let’s make a cloudy sky. I can choose the movie style or preview angle. I really like this one. It looks like light came out of the clouds. Artificial intelligence is scanning the picture and generating a video from it. Now you can see that the clouds are moving, the sky is getting dark, and a storm is coming.

At present, many companies, including Google and Runway, are actively developing such artificial intelligence projects. However, OpenAI emphasizes that "Sora" is outstanding in the realism of its images and its ability to generate video clips longer than other models. However, at present, "Sora" still has some defects, for example, it may confuse some spatial details in text instructions, such as left and right directions.

Artificial intelligence product developer Mostak:I think this is one of the biggest leaps we have made in technology since the emergence of the Internet. It will create a brand-new industry and make the media more exciting and interesting. I think this will inevitably create a lot of new jobs, and I think it will change everything.

However, while artificial intelligence technology brings many new opportunities, it also inevitably brings unprecedented challenges and hidden dangers. Among them, artificial intelligence to generate deceptive content to interfere with elections is considered to be an important challenge facing the world.

On January 23, local time, the Republican primary in the 2024 US presidential election was held in New Hampshire. Prior to this, many American voters said that they had received a "phone call from US President Biden".

False telephone recording generated by artificial intelligence: it’s nonsense. When the votes are counted, we can know the value of democratic voting. You have to save your vote until November.

Soon, White House Press Secretary Pierre clarified that this was a forged telephone recording.

The Italian newspaper La Repubblica reported that the video generation model can make "celebrities or politicians say things they have never said and do things they have never done". If someone uses the model to clone Biden, Trump or others who are running for the US President or the European Parliament in 2024, it may aggravate the spread of false information and false information. Basil Simon, a researcher at Stanford University, worries that these tools will be abused in the election and that the public will "no longer know what to believe".

The video generation model also worries people who make a living by making videos and film practitioners. Valenzuela, founder of Runway Company, bluntly said that the threshold and cost of future film production will be greatly reduced.

Valenzuela, founder of Runway, USA:Using a mobile phone can make a movie watched by hundreds of millions of people, which will change the pattern of the film industry.

Some netizens said frankly: "Fortunately, artificial intelligence video and real video are not too difficult to distinguish at present."

But some netizens are worried: "I’ve had enough. I’m going to lose my job." "Fabricated political propaganda will spread all over the Internet."

Some netizens questioned: "I am curious about the source of learning materials for generating these videos. In order to generate effective videos, it needs a lot of different video data." "The development of artificial intelligence is like a driverless car, and I don’t know where it will go."

Expert interpretation: Generative artificial intelligence technology has made remarkable progress.

The global generative artificial intelligence technology has developed rapidly and is widely used in more and more industries. How to treat the development of artificial intelligence? Let’s listen to Ceng Yi, a researcher at the Institute of Automation, China Academy of Sciences, and an expert from the United Nations high-level advisory body on artificial intelligence.

Ceng Yi, a researcher at the Institute of Automation, China Academy of Sciences, and an expert at the United Nations high-level advisory body on artificial intelligence, said that the progress of artificial intelligence "Sora" in the authenticity and continuity of generated data should be said to be very obvious. The latest development, the cross-modal generation of visual and auditory data, can be said to be by going up one flight of stairs’s user experience.

At present, artificial intelligence still has common sense mistakes.

Ceng Yi, a researcher at the Institute of Automation, China Academy of Sciences, and an expert at the United Nations high-level advisory body on artificial intelligence, said: On the one hand, the research and industry related to artificial intelligence in our country are capable, and we must catch up as soon as possible. Because of "Sora" related generative artificial intelligence technology, it can reduce the cost and threshold of related development of artificial film and television entertainment and cultural products. On the other hand, it also needs to be pointed out that this does not mean that generative artificial intelligence has really understood our physical world, and can completely replace the traditional tasks that need human beings to complete in these industries. For example, the video generated by the current "Sora" system still has many mistakes that violate the common sense of life and physics.

AI’s rapid development supervision faces challenges

Ceng Yi, a researcher at the Institute of Automation, China Academy of Sciences, and an expert at the United Nations high-level advisory body on artificial intelligence: In the future, while actively exploring the potential applications of these technologies and their complementarity to human work, we should pay full attention to the potential negative impact of generative artificial intelligence on society. For example, "Sora" in the continuous high-precision video generation, it uses this technology, so that the level of artificial intelligence fraud has reached an unprecedented height, which brings greater challenges to seeing is believing.

Ceng Yi, a researcher at the Institute of Automation, China Academy of Sciences, and an expert at the United Nations high-level advisory body on artificial intelligence: For the public, it can be said that it is even more difficult to distinguish false information, and the challenges brought by visual forensics and law enforcement can also be said to be very significant. On the one hand, the development and implementation of policies related to artificial intelligence governance must keep up; On the other hand, with the development of generative artificial intelligence technology, its R&D institutions and enterprises should simultaneously develop and popularize a higher level of automated counterfeit identification technology. Only in this way can the generative artificial intelligence technology be developed and used more healthily.

Original title: OpenAI video generation model opportunities and challenges coexist