The explosion of ChatGPT, the release of Baidu's "Wenyan One Heart", Wang Huiwen, Kai-Fu Lee and other well-known figures in the TMT field have entered the game, making the AI field in the bottleneck period of scene application boil again, and AIGC, which is burdened with new value, has also become the focus of heated discussions on the streets.
In this process, as one of the industries that is expected to take the lead in realizing the landing of AIGC, a number of game companies have also started an arms race, through high-salary recruitment of AIGC talents, increased AI technology research and development and platform construction, etc., in order to obtain the ticket to open the era of game AIGC. After the application of AIGC technology greatly improves productivity, game companies will be freed from time-consuming and labor-intensive repetitive work, and "how to use AIGC tools to show game ideas" may become the core issue of game companies in the new round of competition.
In addition, while hoping that AIGC will bring a new round of change to the game industry, risks such as device cost, copyright ownership, and data privacy are still unavoidable and urgent challenges for the game industry.
Game companies grab AIGC
Entering 2023, under the tone of "reducing costs and improving efficiency", game companies are still using relatively generous salaries to recruit talents in AIGC. According to incomplete statistics, game companies such as Tencent, Netease, Sanqi Interactive Entertainment, Giant Network, Perfect World, and Kunlun Wanwei have successively released positions related to AIGC technology, and most of them have monthly salaries of more than 30,000.
When talking about the layout of the AIGC field, the relevant person in charge of Perfect World said that it has promoted the company's game studios to conduct technical training and put into use AIGC. Intelligent integration through AI technology, opening up solutions for the project and meeting functional requirements, such as intelligent NPC in games, scene modeling, AI plot, AI drawing, etc., has played a good role in promoting design ideas and creative inspiration and improving the efficiency of game research and development.
Behind the increase in AIGC by game companies is AIGC's key role in improving game productivity.
The relevant person in charge of Perfect World revealed to the Nandu reporter, "The trend of AIGC has become a reality, if game manufacturers want to maintain unique R&D advantages in the era of AIGC, they must test the team's talent reserve and the ability to apply technology, that is, who uses AIGC tools and how they use them." ”
According to the research report of CSC Securities, games are the most industrialized and interactive/real-time requirements in all content production, and there has always been an impossible triangle of "high quality - low cost - short time". The application of AIGC technology is expected to break the impossible triangle of the game production process, including assisting the game production process to achieve cost reduction and efficiency increase; Upgrade existing gaming experiences; Further lower the production threshold and promote active game innovation.
"The gaming industry has an extremely high failure rate, with only 3 out of 100 games likely to succeed; The initial investment cost is huge, and the R&D cost is often hundreds of millions; In addition, the development cycle is long, and the development cycle of a medium-sized game is expected to be 2-3 years, and it will take longer to obtain revenue. Yin Xueyuan, CEO of Walker AI, said that after using AIGC technology, the game development team only needs to set the initial creative direction, and then hand over most of the time-consuming and laborious execution work to artificial intelligence to form a set of "automated" production lines, thereby greatly reducing the trial and error costs of the game industry.
Wang Xu, co-founder and chief analyst of Gamma Data, said that from the perspective of the job requirements of game companies AIGC, it can be divided into two categories: AIGC art and AIGC algorithm research, the former is to improve the production efficiency of art content, and promote the popularization of related models, and even apply to the user's content production level, reduce the threshold of creation, and accelerate the process of creative realization. The latter is an AI tool for creating and optimizing game content production, meeting the specific needs of collaborative projects in terms of visual, audio, and text generation.
AIGC will also play an important role in improving the gaming experience. Xiao Jian once revealed to the Nandu reporter that you can imagine that the open-world RPG game will use AI to design NPCs, and each NPC encountered is based on a huge setting library of random combinations of interpretations, with its unique background, personality, way of speaking, and truly realizing thousands of faces, the game experience will become very vivid and vivid, and this is only a small part of the possibilities that AIGC brings us.
Relevant information shows that game companies have strengthened their layout in AIGC through internal self-construction and purchase service solutions. For example, the "self-research faction" that personally set up an AI lab or platform includes Tencent's AI Lab and Cros, NetEase's Fuxi and AI Lab, and Mihayou's Anti-Entropy Research Institute; Game companies such as Mu Pu Technology, Longyuan Network, and Lingxi Interactive Entertainment have entered the game by using hyperparameter AI Bot, Walker AI Bot and Qiyuan World Game AI solutions.
According to the "2022 China Game Industry Report", the domestic game market revenue has declined and the scale has been reduced.
From scene applications to participating in content production,
AIGC becomes more "visible"
AIGC is nothing new to the gaming industry. Previously, AIGC has long been applied in the art, battle, plot and other links in the game field, and now AI has made breakthroughs in pre-training large models, text, and image generation technologies, making AIGC technology break through the limitations of intelligence and versatility, and appear in front of the public with a more visible attitude.
Some insiders said that in the past, AI played a large role in separate categories or individual functions, and intelligent NPCs in MOBA and FPS could have the competitive level of top players. For example, DeepMind's AI program "AlphaStar" defeated two top StarCraft 2 pros with a score of 10:1.
The reason why AIGC has a higher "degree of display" is because it involves not only the application of AI technology in the game industry, but also the fundamental transformation of AI technology to the game industry. Yin Xueyuan said that reinforcement learning technology similar to AlphaGo is mainly used for game NPC, game testing, numerical optimization, etc., which is a good application scenario but does not solve the fundamental problem. AIGC, on the other hand, allows AI technology to directly participate in the generation of game content. This means that game developers can use AI technology to generate characters, scenes and other content in the game, thereby accelerating the progress of game development, improving game quality, and reducing game development costs.
Wang Chuanpeng, senior technical director of 37 Interactive Entertainment Group, said that compared with the previous wave of AI boom (2016-2020), the key update point this time lies in the breakthrough of pre-trained models, various generation technology branches and cross-modal technologies, so that AI has been optimized in terms of intelligence, application threshold, versatility, etc., which is expected to bring scalable business value.
Tencent's mobile game "Metal Slugg: Awakening" received a version number.
The highly anticipated domestic 3A masterpiece - "Black Myth: Wukong".
The game ecology has quietly changed,
Creative expression of ingredient watershed
With the deepening of AIGC's application in the game industry, the traditional game development model with long cycle, high cost, low efficiency, and limited creativity may no longer plague game development, and the game industry ecology and division of labor will also change.
Wang Chuanpeng said that some pipeline processes in the game industry are likely to be greatly improved by AIGC talents; The game industry will enter the competition between man and machine and another group of people and machines, and how to make good use of AI may become the core competitiveness of the development of game companies.
At the same time, changes in the types of jobs and their job responsibilities in the game industry have also become an unavoidable topic in the game industry.
"Chatting with two game teams, team one cut the original art outsourcing company, and the second team cut the translation outsourcing company. AI is already actually starting to affect many jobs. Huang Yimeng, CEO of Xindong Company, once said, "AI may not be able to replace you, but in the future, people who are good at using AI as role models can replace you." Everyone needs to be ready to embrace change, use their original professional skills and accumulation, and use AI as a tool for their own use at the first time. ”
When talking about AIGC's role in game empowerment, Guo Weiwei, CEO of Xishanju, believes that "the future career orientation of original artists may become original art trainers, who used to draw by themselves, and may tell AI what to draw and continue to learn and correct." ”
"At present, the change in the division of labor in the game industry chain is not obvious, because AIGC cannot replace the outsourcing of art and translation from the perspective of usability and economy for the time being." However, Analysys analyst Liao Xuhua said that AI has already had an impact on the content division of the game industry. For example, TA (Technical Art), which has emerged in the past two years, is an engineer who combines technology and art based on AIGC, and TA also leads the collaboration between game art and program in the project.
In this process, "how to use AIGC tools to present game ideas" may become the core topic of game companies in the new round of competition.
Yin Xueyuan said that after the application of AIGC technology greatly improves productivity, it will make the game market more diversified and personalized, and game companies need to pay more attention to user experience and game creativity to create more attractive and distinctive game products.
"AIGC is just a tool for game development, which is essentially the same as engines, data analysis, motion capture, etc., it's just a tool." Liao Xuhua believes that the essence of the game is still creative work, because AI does not have emotional thinking, so AI can not accurately understand the emotional content to be expressed by the game, plan to put forward a character original painting needs, excellent designers can understand the character's personality, story through communication, and draw it well, AI can not replace this process, can only become a designer's tool in the drawing process.
In addition, Yin Xueyuan also analyzed that the application of AIGC technology will also give birth to new game markets and business models, such as game development platforms based on AI technology, game customization services based on AI technology, etc., which will bring a richer and more diversified game ecology.
Cost, copyright are still obstacles,
But the "killer" app is on the horizon
"At present, AIGC is mainly used in text and images such as graphic production, sound synthesis, and text output, which is limited by technical maturity and is only in the exploration stage in 3D content and video." Wang Chuanpeng said that at this stage, the 3D models and video materials used in game production and distribution are still difficult to generate AIGC, and there is still a gap from the industrialization level, because this process is very complicated, and there are still technical challenges for AIGC generation. In terms of natural language, ChatGPT also has a gap in correctly solving user problems, and it requires an open model combined with company data for small-scale adaptation to solve this problem.
Therefore, AIGC generation of 3D models (including textures, bones, and motions), AIGC generation of video (frame streaming technology), and AIGC generation of dialogue that meets the settings (intelligent customer service) are still "make-up lessons" for AIGC.
At the same time, how to calculate the cost account and avoid property rights risks is also the key to whether AIGC can successfully land and develop in the game industry.
Wang Xu said that first, the research of AI technology requires a lot of resources and costs, and many game companies, especially small and medium-sized teams, are more difficult to develop AI tools by themselves, and can only use industry general products for further research, which will involve many problems and take longer to popularize. The second is how existing talents can truly accept and control AI tools, especially for positions that require creative work, how to use AIGC to improve efficiency.
"The copyright risks of AIGC are particularly concerning." Some game industry analysts frankly said that at present, the mainland legal system has not clearly divided the copyright ownership of AIGC, and further discussion and division of rights and responsibilities are still needed on whether AIGC constitutes a work, who should belong to AIGC works, and whether AIGC will involve infringement.
The relevant person in charge of Perfect World said, "At present, the development and application of AIGC technology in the industry is still in the testing period, and there is still a broad space for development in the future. ”
In this regard, Liao Xuhua added that at present, the application of AIGC in the game industry is mainly experimental, and there are few real large-scale and normal applications, and technical reliability and economy still need to be improved. The next stage is to start trying large-scale applications one by one, comprehensively apply them in a certain product or link, and constantly adjust them with trial and error, and finally form a relatively stable and extensive application stage, like a game engine, which has become one of the basic tools for game development.
Sequoia China predicts in Generative AI: A Creative New World that text and code AI generation will enter a mature period in 2023, image AI generation will enter a mature period in 2025, and video, 3D, and game AI generation will enter a mature period in 2030.
At the same time, Sequoia China also said in the above article that AIGC is now in the fourth wave of development, killer applications appear, with the stability of the platform layer, models continue to become better, faster and cheaper, the acquisition of models tends to be free and open source, and the creativity of the application layer has matured. "Just as the inflection point of mobile internet was opened up by some killer apps a decade ago, we expect killer apps for generative AI to emerge and the race begins."
Written by: Nandu Bay Finance Agency reporter Chen Peijun