In the past two years, virtual digital humans have attracted the attention of many people along with the metaverse trend. In China, virtual digital humans have begun to be used in some fields, such as live broadcast delivery, customer service, etc. Although virtual digital humans seem to have good prospects, in fact, there is still a long way to go before virtual digital humans can be fully implemented, and there are some challenges that need to be overcome.
technical level
The first is the technical challenge. Affected by technology, it is actually difficult for digital humans to be infinitely close to the real nature of humans, because AI technology is still in the early stages of development. Therefore, although a digital human looks like a human on the outside, its core is far from the state of a real person, or close to the state of a real person. For example, digital humans still lack a lot of affinity in terms of their expression management and perception interaction.
Although many movies such as "Ready Player One" and "Blade Runner 2049" have brought a lot of imagination and expectations to virtual digital humans; however, the reality is that whether it is metaverse platforms such as Meta and Hee Rang, or various All kinds of virtual digital humans can provide users with quite limited experience. Even the virtual digital person with the head has some complaints about being "ugly" and "stupid". Perhaps it is because the movies are too realistic and realistic, which makes today's virtual digital humans unable to give the users of the Metaverse too much surprise.
Liu Wei, founder of Hualian Digital Technology (a domestic technology company that mainly builds a comprehensive service platform based on digital human technology + content + operation), believes that rapid generation and AI drive are the two major technical difficulties in the current virtual digital human industry. Only by realizing the rapid generation of virtual digital humans, solving the two major problems of low-cost mass replication and high-frequency content output, and getting rid of the shackles of middle-aged people, can virtual digital humans have universal commercial value. However, the reality is that the production cost of high-quality virtual digital humans is extremely high, the production cycle is extremely long, and the effect cannot be guaranteed. At this stage, using virtual digital humans to create content is a very risky thing.
Since 2022, companies including Douyin and Kuaishou have successively released virtual digital human-related technologies and support plans. However, only when the virtual digital human has a sense of reality and immersion comparable to that of a real person, and at the same time, the cost is lower than that of a real person, will the virtual digital human have more general commercial value and be more widely accepted by the public.
There is still a big gap between virtual digital humans and virtual digital Homo sapiens. Most virtual digital humans stay in a realistic shell, and the routine response is mainly based on basic settings. The technology that combines IQ and EQ is still being explored. Many virtual digital humans are like NPCs in the game, unable to interact with the audience. Then there is the high cost of high-precision motion capture and the long post-development process. For example, the virtual digital human rendering process of the 2019 American science fiction movie "Alita: Battle Angel" took 30,000 computers, 800 technicians, and a total of 432 million hours to complete.
On the whole, the overall market value of virtual digital humans actually depends on the advancement of technologies related to virtual digital humans. In the future, virtual digital humans will usher in a rapid growth path similar to the AR/VR industry, and they will all go through a long period of technological advancement. Only time will give us the answer.
product level
For now, many virtual digital humans are not differentiating. At present, many virtual digital human companies are developing and applying based on open source technologies such as UE5. Few companies have overwhelming technical advantages, and most of them can only "change the soup without changing the medicine" in terms of content, operation, and creativity. Wang Boxuan, founder of Quantum Craftsman (a domestic company that focuses on creating virtual digital humans), believes that the biggest problem with virtual digital humans at present is that they are not deep enough and have insufficient value recognition. Turning pictures into virtual digital humans is just an expensive show that doesn't make sense in itself.
Then there is the shortage of talents. Since virtual digital humans are currently a brand-new track, it is difficult to supply talents in the market, and the lack of talent accumulation has created a big talent gap. The shortage of talents has also made it difficult to produce relatively good products on the market, because there are not enough talents to participate in the design and development process of virtual digital humans.
Just like the text-based official account and the video-based Douyin require completely different content gameplay, as a new content form, the virtual digital human must match the new content gameplay. However, the reality is that the current gameplay of virtual digital humans is more to humanize virtual digital anchors, entertainment anchors, and traffic idols, and has not yet truly released the content value of virtual digital humans.
market level
cost issue. The cost of making a virtual digital human is very high. The insight research report shows that the unit price of virtual digital human production is more than one million yuan, and the cost of higher-precision virtual digital human can even reach tens of millions. Taking Douyin beauty expert Liu Yexi as an example, Liang Zikang, CEO of Chuangyi Technology, the production company, said that Liu Yexi’s production investment alone is at the level of one million yuan, and the first short video of Liu Yexi costs about a few dollars. One hundred thousand yuan. The cost per second of ultra-realistic virtual digital human videos is in the range of 10,000 yuan. It can be seen that the initial investment cost of virtual digital human is extremely high. From the initial design to 3D modeling, to the investment of AI technology, the virtual digital human has very high requirements for every link, and each link actually requires a lot of money (in fact, every link is now burning money. stage, especially for investment in AI technology). In addition, the cost of digital human production mainly depends on the breakthrough of AI algorithm. Breakthroughs in AI algorithms can significantly help reduce the production cost of virtual digital humans.
At present, the application of virtual digital humans at the market level is still insufficient. For example, virtual digital humans currently prefer entertainment consumption and virtual customer service application scenarios. For example, "Jia Le", the first virtual anchor at station B, once set a monthly sales record of 2.14 million yuan, and a live broadcast of a birthday party hosted by him set a record A sales record of 1.89 million yuan. The application of virtual digital humans in other markets such as education, medical care, elderly care and other public services is relatively weak. The current situation is that the results of scientific and technological innovation have not served a wider range of people, but only a small number of people.
The virtual digital human industry still lacks a comprehensive definition and unified standards, so the development of the entire industry is still in a stage where good and bad are mixed, and a hundred flowers are blooming. From this point of view, whether the era of one digital person per capita can really come, not only considers the powerful AI technology of major manufacturers, but also needs to create a more complete ecology.
legal aspect
With the further advancement of the metaverse in the future, the difference between the real world and the virtual world will become more and more "distinguishable". The virtual world is easy to cause people to have the illusion of identity, resulting in the emergence of a "grey area" of ethical order in the use of virtual digital humans, causing risks of "face-changing", "immortality" and "morality", posing higher challenges to the ethical order. At the same time, because the virtual digital person is not a legal subject, the ownership of its image IP, story IP, and brand IP is still in the legal blank, which may easily lead to potential risks such as copyright disputes, privacy leakage, and data security. For example, Ada, the first "surreal" AI artist, can complete unique artistic creations, but the copyright ownership of these works cannot be determined.
In addition, if someone wants to marry a virtual digital person, can the existing law recognize it? In fact, Japan already has an example. In 2018, a young Japanese man married Hatsune Miku, a well-known virtual idol, unofficially. Friends and family of the man expressed disapproval. With the rapid development of technology and the continuous advancement of virtual digital humans, more and more people will choose to "marry" virtual digital humans in the future, so how will society accept this group in the future? And how should the law be amended? These will be a problem that virtual digital humans will encounter when they land in the future.
How the government will regulate virtual digital humans in the future will also be a problem. Microsoft once developed an AI and posted it online for training. However, under the operation of some people with serious intentions, the AI actually talked about racial discrimination and other words. After learning about it, Microsoft also urgently removed the AI. How the government conducts a certain degree of supervision and the unclear regulatory authority of the government will also be a problem. In the future, virtual digital humans will play multiple roles, and regulatory agencies will involve different agencies such as culture, education, public security, industry and commerce, and telecommunications, resulting in overlapping or missing regulatory scope. At present, the regulatory standards of most virtual digital human companies are still being explored, and a unified standard has not yet been formed, resulting in risks such as technological regulatory loopholes and loss of control of public opinion management.