AI has already come out of the laboratory and is 'on the job' in game content production processes such as drawing, text, and audio after only one year. Wang Chuanpeng, Vice President of Data at 37 Interactive Entertainment Network Technology Group, told Caijing News that General Office AI 'Xiaoqi', professional AI Agent platform, and optimization of AI in user experience are the company's main directions in AI research and development. 90% of employees in the company have integrated AI into their work. In terms of product structure, the operating income of 10-30 billion yuan products is 401/1288/60 million yuan respectively.
On August 2, Caijing News (Reporter Cui Ming) reported that artificial intelligence is accelerating the change in various industries. In 2023, there will be 0.06 million open-source AI projects released on the GitHub community, which is 3.5 times more than in 2022.
When the generative AI wave hit last year, the gaming industry became the first to perceive and adapt to the impact of AI. The experimental results and application prospects of 'game + AI' have also become a hot topic in the industry. In just one year, AI has already come out of the laboratory and is 'on the job' in game content production processes such as drawing, text, and audio.
'This year, there are many practical applications in the entire industry, and everyone truly feels their value. There will be more applications landing in the future.' During ChinaJoy this year, Wang Chuanpeng, Vice President of Data at 37 Interactive Entertainment Network Technology Group, said in an interview with Caijing News that General Office AI 'Xiaoqi', professional AI Agent platform, and optimization of AI in user experience are the company's three main directions in AI research and development.
Regarding the strategy of layout, Wang Chuanpeng said to start with the company's internal needs, closely align with business needs, and empower internal teams. After the internal AI pipeline matures, the possibility of commercializing it and pushing it to the external market cannot be ruled out in the future.
Change starts from within.
Currently, 90% of 37 Interactive Entertainment Network Technology Group's AI applications are concentrated within the company to serve internal team business needs and help employees improve efficiency.
These AI Agents used internally mainly fall into two categories. One is General Office AI 'Xiaoqi', currently used to improve employees' daily office efficiency, such as information consulting, application entrance, copywriting editing, translation, drawing, etc.
Wang Chuanpeng revealed that most employees are using 'Xiaoqi', and the daily call volume per person has reached more than ten times, mostly used for checking code, parsing documents, information searching, and translation content.
The other is a professional AI Agent platform used to support business staff in professional office work, such as 2D drawing, video production, data analysis, art resource expansion, tool and process-type intelligent agents.
Under the help of professional intelligent agents, Wang Chuanpeng revealed that in terms of art, the current output can reach 0.3 million images or even more in a week, and the output effect can fully meet the needs of different teams in research and development and distribution. One of the offline customer service teams under the internal distribution line only spent 900 yuan to complete basic communication tasks that required more than 20 customer service personnel in the past, saving customer service manpower for more difficult work.
Currently, whether it is overseas companies such as Microsoft, Google, and Meta, or domestic technology giants such as Alibaba and Baidu, they all regard AI Agents as an important hand for the application of large models in various industries and have launched various AI Agent products.
Wang Chuanpeng told Caijing News that there are differences between 37 Interactive Entertainment Network Technology Group's AI Agents and those of these technology giants. 'First, this system is self-developed, and a lot of optimization development has been done for users and Agent planners. Secondly, it delves deep into the gaming business process. All of our Agents can correspond to our business processes, and they all come from business experience and needs. Finally, we have formed an iterative effect algorithm flywheel model.'
Regarding whether to consider commercializing these internal AI capabilities, Wang Chuanpeng said that after the internal AI pipeline matures, there is no denying the possibility of commercializing and pushing it to the external market in the future. 'But this is not our current priority, and the core is to empower internal teams first.'
Optimizing user experience is the next step.
In building the 'AI Office Middleware', 37 Interactive Entertainment Network Technology Group is also exploring more possibilities of 'game + AI' in building new user experience.
In this direction, Wang Chuanpeng said that the team has completed the first phase: the verification of game AI's full elements generation and is currently carrying out the optimization of AI in user experience.
In the future, where will AI go in the gaming industry? Wang Chuanpeng believes that, firstly, productivity will be further accelerated. Resource production, code production, and planning will all have AI applications. Secondly, because of AI, certain suitable categories (such as simulating management, Team RPG, etc.) enhance users' sense of participation, and the game world becomes more open and flexible. UGC and recommendation mechanisms will be introduced, allowing game users to feel a better gaming experience with the help of AI.
'Because games have landed, AI will have more data and models for training, and its ability will become stronger. AGI should be possible with the addition of games.' Wang Chuanpeng said.
Regarding the company's investment focus, Wang Chuanpeng said that both value and difficulty need to be considered. For example, translation is of high value and difficulty. Especially for companies that publish games overseas, localized translation is a significant cost. They have enough technology ability now, so the team will spend energy on doing this.
In the fields of video and 3D, it is still in the exploration stage. "Solving the problem of video and 3D generation can greatly enhance the efficiency of research and development and publishing, and also improve the user experience. However, the current implementation difficulty is very high, and the company is contacting external models while also solving some internal difficulties," said Wang Chuanpeng.
Wang Chuanpeng revealed that AI is a "top priority" project at 37 Interactive Entertainment Network Technology Group. The manpower investment in AI-related projects is basically more than 50% of the overall manpower of the technical center, and almost all closed-source and open-source large models on the market have been accessed. In addition to independent exploration and internal incubation, the company also uses external investment to further cut into the AI field. Investment in AI has been elevated to a strategic level.