February**Dynamic Incentive Plan Author |Maggie.
Edit |Liu Jingfeng.
A year ago, the scene of the emergence of AI intelligence has reappeared.
At the beginning of 2024, SORA has triggered a collective restlessness in the global tech community.
On February 15, local time, OpenAI released the latest ** generative model SORA. According to OpenAI, Sora can generate up to 1 minute of HD** based on short text instructions entered by the user and create complex scenes with multiple characters, specific types of movement, and accurate subject and background details.
What amazes the global audience is the very realistic picture presentation in SORA generation**, and even has a little cinematic texture.
Source: Screenshot of openai**.
In a series of different themes** released by OpenAI, whether it is the scene of a woman wearing sunglasses walking on the streets of Tokyo, or the detailed shaping of animated characters, although these contents are completely generated by AI, the realism of the details has entered a new dimension. This has made many netizens lament that the boundaries of reality are becoming more and more blurred in the digital world. As OpenAI proudly writes in the title of SORA's technical report, "Generative models like SORA are simulators of the world
For Chinese enterprises going overseas, the significant improvement in content productivity driven by AI technology will also further help industrial upgrading and accelerate the pace of globalization of enterprises. And the TikTok platform has begun to be filled with all kinds of AI-generated **crazy fans.
At the beginning of last year, ChatGPT, a large language model released by OpenAI, set off a boom in artificial intelligence. After ChatGPT was launched, almost all global technology companies have thrown themselves into the AI field at full speed and released their own large language models, which has affected the changes in the way people search for information and find materials in their daily lives.
The release of the Wensheng model SORA more intuitively shows at what speed AI technology is developing rapidly.
Under a SORA demo on social **X**X, Tesla founder Elon Musk commented that "GG Human" (GG is the abbreviation of good Games, which means "play well, I admit defeat"). He also said that humans, enhanced by artificial intelligence, will produce the most outstanding works in the coming years.
Zhou Hongyi, founder and chairman of 360 Group, said that the birth of SORA means that the realization of AGI will be shortened from 10 years to 1 year.
SORA pushes AI capabilities to a new stage and is the beginning of practical use. **Production costs will be significantly reduced, and the skills required will be completely different than before. Previous skills are how to find a quality shooting team, location, and the right product for the shoot, how to coordinate and communicate. After that, the skills are how to write prompts, how to let AI integrate their own products into ** without distortion, and how to make ** look real, clear and textured. Guo Lin, founder of passioncy, told Xiaguang Society.
Just recently, on OpenAI's TikTok account, all kinds of AI-generated ** crazy fans, attracted more than 100,000 fans in just a few days, and received more than 500,000 likes. Netizens have left messages lamenting whether the creators of short ** will be robbed of their jobs by AI, and even the TikTok platform will be affected.
Although SORA has not yet been officially opened to users, it has already made waves in the capital market. The market expects OpenAI to be valued at more than $80 billion in the latest round of funding led by venture capital firm ThriveCapital. For comparison, OpenAI was valued at about $29 billion shortly after it released ChatGPT early last year.
Source: Unsplash
The day after SORA's release, the stock prices of several U.S. technology companies fell in response. According to ** reports, the stock price of adobe, an American computer software company, exceeded 7%; Shutterstock, a U.S.-based library, material, and editing tool provider, fell more than 5%; Shares of Google's parent company, which released the "Wensheng**" tool Lumière a few weeks ago, fell 158%。In one day, the combined market value of these three companies evaporated by nearly 48 billion US dollars (equivalent to nearly 350 billion yuan).
For many industries, SORA is like a "bombshell", making practitioners both excited and nervous as they look to the future.
Among them, the film and television industry bore the brunt. In the traditional film and television production process, post-production and special effects production often require a lot of manpower, material resources and time, and AI tools such as SORA will greatly reduce these costs, helping film and television production companies quickly create high-quality first-class clips and shorten the production cycle.
At the Berlin Film Festival, SORA became one of the hotly discussed topics among filmmakers.
According to foreign media reports, Los Angeles director Dave Clark (D**e Clark) believes that creators need to embrace artificial intelligence technology to make content that has not yet been imagined or realized, rather than feeling threatened.
Christina Caspers-Roemer, managing director of German VFX studio Trixter, said AI tools like SORA are proving to be more efficient and faster in workflows, but even so, film and TV production will still be based on human creation. "Eventually, our customers will always come back to the real world. She said.
From ChatGPT, DALL-E3, to SORA, AI content creation has swept from the creation field to the creation field.
In fact, SORA is not the first Wensheng ** model to receive attention. In recent years, the world's major technology companies have almost invariably invested heavily in the field of artificial intelligence, and plan to continue or even increase investment in the future.
Before SORA was officially unveiled, most of the world's leading technology companies already had their own Wensheng ** large model.
In November last year, the Wensheng ** model ushered in a wave of explosion. On November 3rd, Runway released the Gen-2 update, which supports 4K definition works; On November 16, Meta released Wensheng's ** large model emu video; On November 18, ByteDance released Pixeldance; On November 21, Stability AI released the Stable Video Diffusion ......, a generative model
In January of this year, Google also released a generative model, Lumière, specifically designed to convert text to.
Currently, SORA is still in the early stages of development, and OpenAI says that the current model still has a lot of weaknesses. SORA may have difficulty accurately simulating the physics of complex scenarios, and may not be able to understand specific instances of cause and effect. For example, a person may take a bite of a cookie, but afterward, there may be no bite marks on the cookie.
In addition, the model can confound the spatial details of cues, such as confusing left and right, and may have difficulty accurately describing events over time.
And careful netizens also found a lot of bugs in Sora's demo**. "For example, in a ** generated by SORA, the scene of the dragon dance in the Chinese New Year is presented, although the text on the dragon and the building in ** is very similar to the font written in Chinese calligraphy, but it is actually not Chinese characters, no one knows what is written, it feels like it is just Chinese imagined by AI. The cat in the other ** has three front feet when it moves. A reader who pays attention to the AI industry told Xiaguang Society.
Source: Screenshot of openai**.
But even so, regardless of the technical aspects such as the generation time, the realization of a single multi-camera, and the real world with accurate physical rules, SORA has achieved a "crushing" lead over other Wensheng models, promoting AI generation into a new era.
Generative AI tools like SORA have significant advantages in reducing costs and improving efficiency, bringing unprecedented development opportunities to industries such as content creation, advertising and entertainment, and also allowing different types of Chinese enterprises to see opportunities.
Previously, some companies applied ChatGPT to multiple scenarios to reduce costs and increase efficiency. According to a survey by AMZ123, 33% of cross-border companies will use ChatGPT in 2023, and nearly 15% of companies are expected to use it.
Nowadays, SORA is not yet fully open for use, but it is conceivable that through more accurate generative AI, whether it is making platforms such as TikTok, or making introductions and marketing for your own products, it will greatly reduce costs and improve efficiency.
How to use AI to produce the best that can really be implemented is something that everyone is beginning to explore. SORA has brought a disruptive impact to the industry, and whoever uses new AI technologies, including SORA, first, will be able to enjoy the first wave of industry dividends. Guo Lin, founder of Passioncy, saidChinese companies must make full use of their existing AI capabilities when going overseas。For small and medium-sized enterprises, especially many enterprises integrating industry and trade, the difficulty is actually very high. That's one direction we're looking to help with right now. ”
A veteran of the advertising industry also told Xiaguang that she believes SORA will disrupt the content of the travel, lifestyle and service industries. As the production process becomes simpler, the boundaries between big and small brand ads will gradually narrow, and content creative links such as "storytelling" will once again become the most important factor in advertising.
Source: Unsplash
In the field of game going overseas, Ada Liu, founder and CEO of Share Creators, believes that the changes brought by ** generation will be "cross-era".
In the gaming industry, AI can help generate upfront concept maps, UI icons, etc. The process of computer graphics production is very long, from 3D production, to rendering, compositing, each link requires a large number of professional personnel. Especially in the style exploration stage, if the ** is directly produced, it is equivalent to skipping directly to the last step, saving a lot of production time and cost. ”
However, Ada Liu also believes that there is still a long way to go for AI-generated content to completely replace humans. "Maybe in another year or two, it can replace some low-end batch production jobs in the 2D part. However, artists who set the direction such as the main beauty will not be replaced. After all, the art style of a product needs to be chosen by people. ”
According to ADA, Sora's demo** is impressive, but it's important to focus on the actual results, as it's not possible to try it out at the moment, so we'll have to wait and see. Guo Lin also believes that the main gap between AI tools and the output of "real landing" is still in the authenticity of content and the ease of use of tools.
According to IDC**, in the next five years, AIGC will have a huge impact on marketing, software testing and other industries. By 2025, 35% of businesses will have mastered the use of generative AI to develop digital products and services. By 2026, generative AI will take on 42% of traditional marketing chores such as search engine optimization, content and optimization, customer data analysis, segmentation, lead scoring, and hyper-personalization. By 2028, generative AI-based tools will be able to write 80% of software testing work, reducing the need for manual testing and improving test coverage, software usability, and quality.
It is foreseeable that even if AI-generated content still faces potential ethical and abusive risks, its unlimited potential for creativity and efficiency will continue to attract increased investment and competition from global tech giants.