In the first month of 2024, OpenAI released the Wensheng ** tool - SORA, which generates ** texture that is amazing, quoting the words of the big V in the technology industry: I didn't believe it was true before, but now I don't believe it is fake. What is SORA? What's different with previous AI tools? What are the implications for social development? Aroused heated discussions. (This article is not written by AI and is hereby declared.) ）
What is SORA? What are the features?
In recent years, the development of AI and large-scale model-related fields can be described as one step at a time, step by step. Starting from speech recognition text, to ChatGPT3, ChatGPT4 text interaction tools, as well as AI mapping tools such as Midjourney, to now VR interactive devices such as Vison Pro and the Wensheng ** tool SORA that we are going to discuss today. The power of SORA is not just the simple generation of text generalizations, but the generation of ** through program architecture, through continuous training. For example, in the past, when making 3D**, each frame was modeled by a 3D designer and then rendered by a calculator, and the production of each hair strand and every micro-expression required a large amount of engineering, and the effect was completely controlled by people. In addition, the most incredible thing about SORA is that the virtual ** it generates is in line with the laws of real-world physics. The founder of Runmi Consulting @刘润 made an analogy, if you bite into food, there should be a bite mark on the food, which is a "physical law", if the food is still complete after biting, it does not conform to the "physical law". The sense of space, gravity, and so on are all physical laws of the human real world. Therefore, many technology bloggers believe that SORA is building a "world model" and is a virtual replica of the real world of human beings. Sora once means "endless" in Japanese, and it is worth thinking about why OpenAI gave it this name.
What industries will the emergence of SORA have an impact on?
At present, the impact on all industries related to images is relatively intuitive. The release of SORA led to the response of Adobe, a leading image processing company41%。Some analysts on the Internet believe that the first is the film and television industry. In the current film and television industry, from stage art, makeup, props to lighting, videography, photography, from actors, screenwriters, drama affairs to animation special effects, if AI ** follows the current development speed, AI can be used to complete. The second is the advertising industry. Some bloggers ridiculed: In the future, advertisers KV (visual main screen) will rely on Midjourney, copywriting will rely on ChatGPT, and TVC (TV commercials) will rely on Sora. In the future, large advertisements such as cosmetics and automobiles will save a lot of advertising costs. The third is the short industry. Whether it's Douyin, Kuaishou, Bilibili, Xiaohongshu, or TikTok abroad, Sora can now generate 60 seconds**, whether you want to make a mosquito fly from the Earth to Mars, or make a sci-fi picture of a submarine sailing in human veins, you only need a command. AI will bring a revolution to the current short-term industry, and it will be difficult for practitioners to cope with this wave if they lack creativity. Fourth, university education. Although university education is not considered an industry, it is broader and more influential. For example, the major of "film and television photography and production" alone is offered in 87 colleges and universities across the country. And there are more similar majors such as "multi-** design", "digital ** art" and "animation design". And it is possible that many of the younger brothers and sisters here have studied 3D character design for four years, and when they graduate, they may not be better than a primary school student who can command AI. In the future, AI + healthcare, AI + social networking, and AI + education will all have an impact on traditional industries.
Are we ready for the development of AI?
The concept and algorithm of SORA was first proposed by Google, a giant enterprise of Google, which was a big company and was under the pressure of industrial monetization, and finally failed to take shape in Google. Therefore, the innovation and development of AI is constrained by many factors such as capital and ecology. If China wants to develop AI, it needs to clarify the best problem? The comprehensive online information is mainly as follows, model: original, assembled or cascading? At present, among the domestic large models, they are mainly divided into three categories: first, original large models; the second is the shell of foreign open source models; The third is to assemble large models, that is, to put together the small models in the past and turn them into "large models" with large parameters. Among them, the number of original large models is the smallest. However, the current situation is that most of the participants are "rolling" the so-called "original large models", and they are worried that the risk is too high, so there are a large number of large models that are shelled and assembled. Hashrate: stuck neck or don't want to buy? Despite the "export control rules" of the United States, it is not impossible for Chinese companies to obtain NVIDIA's computing power, and there are many ways to choose from it. The GPU itself is very expensive, and the deployment, operation, debugging, and use after buying it are all costs. Previously, there was a saying in the industry that many scientific research institutions in China could not even afford to pay the electricity bill for A100. Data: How to solve low-quality data? The Chinese data used by openai to train large models is also the same as that of Chinese Internet platforms, but it has done a lot of extra work to improve data quality, which is not something that ordinary data labeling work can do, and requires a professional team to clean and sort out the data. Capital: Is only capital short-sighted? Although OpenAI has blazed a clear path, for the vast majority of companies, the cost and time required to make a mature large model from scratch is not much shorter. The patience of Chinese investors is only three to five years, which is determined by the mode of capital operation. (The above views are from @Caijingeleven).
What will be the impact of SORA on society in the future?
First, it will draw a large amount of social resources. AI is now seen as an important enabler of social development, comparable to the Industrial Revolution. Apple's VisonPro is priced at 25,000 yuan per unit and supports installment payments. This is only a division-level application in the field of AI, and the future market prospect can be imagined. Sam Altman announced that it will raise $7 trillion to continue to invest in the development of OpenAI. Such a large financing scale is comparable to the market value of all Internet companies in the world, and a large amount of funds will flow to OpenAI, which will attract and gather a large number of high-end talents.
Second, it may have an impact on the existing social form. Mao Xingliang, a professor at Hunan University of Technology and Business, believes that the development of AI will allow the capitalist class to gradually get rid of its dependence on the proletariat. He said that in a capitalist society, capitalists need to exploit the physical and intellectual strength of a large number of proletariats in order to become capitalists. With the advent of AI, a set of equipment and a batch of algorithms can be realized, and there is no need to squeeze human intelligence, and the proletariat will become a "useless class", which will lead to the loss of human value and the natural decline of the population. To sum up, the emergence of AI will usher in new models and challenges for social management.
Third, the development trend of AI is gradually worrying. There is a paragraph on the Internet**: Some bloggers asked, to what extent will people and AI develop in the future, and what balance should be maintained between each other? AI answers that AI will live in harmony with humans. This deceptively simple answer is chilling. We usually say that when humans and nature coexist in harmony, people are the main and key elements, and people should be put first, but the answer of AI puts itself in the main position. Illustrating that AI may have emerged as self-aware. Nvidia founder Jensen Huang believes that the global AI data center will double in the next 4-5 years. Zhou Hongyi, founder of 360**, general artificial intelligence AGI will also arrive in two or three years. At present, virtual reality applications such as brain-computer imaging and VisonPro seem to be passively serving AI.
In the future, algorithms will continue to improve, data will grow exponentially, and the social effects of AI will continue to change. How to make AI develop in a healthy and orderly manner may be more necessary than AI itself. AI ethics and compliance and the improvement of relevant laws need to be planned in advance.