Interview with 37 Interactive Entertainment: What can AI really achieve now?

Text by: Dried Fruit

Currently, the AI workflow is a matter of great concern to many companies. Simply put, it refers to optimizing a company's overall management, production, communication and other processes through AI, significantly reducing the company's internal costs, and ultimately enabling the company to outperform traditional enterprises and achieve groundbreaking results more quickly.

When it comes to specific operations, many companies are still in the exploratory stage and are even relatively cautious. For example, in project management, with costs reaching millions or even tens of millions, no one dares to blindly trust the judgments given by AI. And when it comes to work hand - over, even if AI provides comprehensive content, many people remain skeptical and are even willing to spend some time comparing the original content. In short, AI cannot take responsibility, so it's difficult for people in companies to fully trust AI.

However, the help that AI can provide is also obvious. For instance, it can assist in meetings, optimize PPTs, or generate various types of low - cost content. If these capabilities are utilized effectively, companies can often avoid many unnecessary costs and even overtake their peers in an instant.

Therefore, regarding the issue of the AI workflow, although many companies know it is a "battleground," they don't know what attitude to adopt or which specific direction to target.

In this regard, we didn't directly use AI, but we did interact with many AI - related practitioners and heard their different views as industry insiders.

Regarding company management, not long ago, we had an in - depth discussion with Wang Chuanpeng, the vice president of data at the technology center of 37 Interactive Entertainment. As one of the earliest companies in the gaming industry to deploy AI, they pay great attention to the help that AI can provide in creative and workflow optimization, and they entered the field earlier.

Wang Chuanpeng, Vice President of Data at the Technology Center of 37 Interactive Entertainment

01

In terms of AI - assisted work, 37 Interactive Entertainment is currently at the forefront of the domestic gaming industry.

As early as 2018, 37 Interactive Entertainment began to deploy in the field of artificial intelligence and invested in researching related content. For example, in 2022, they focused on research in the field of natural language and computer vision (CV) and increased their investment in computing power. After the popularity of GPT the following year, they further increased their investment in technology procurement (including computing power, human resources, etc.). Up to now, the company's investment in the AI field has been gradually increasing.

The results are quite obvious. When AI became popular and some people just started to notice the potential changes it could bring, 37 Interactive Entertainment had already implemented a mature AI application solution and launched an independently developed industry - specific large - model named "Xiaoqi" within the company.

In other words, "Xiaoqi" is a direct manifestation of 37 Interactive Entertainment's exploration of AI. Currently, it can be applied to game development, distribution, and many general office scenarios.

Taken at the ChinaJoy venue

Take the most commonly used meeting minutes function in the office field as an example. The AI meeting minutes agent in "Xiaoqi" can accurately record meeting content and achieve functions such as one - click transcription and intelligent analysis. Usually, within 5 minutes after a meeting ends, it can provide a highly complete meeting minutes, saving a series of complex operations such as recording, uploading, and translation.

Of course, "Xiaoqi" can do even more. It has currently built more than 40 types of AI capabilities and applied workflow optimization to all business segments of the company. For example, all games under the company have been connected to AI customer service, including "Douluo Continent: Soul Hunting World" which was launched recently. Wang Chuanpeng said that "the accuracy of the AI customer service has reached 80%, equivalent to the work efficiency of a 30 - person customer service team."

Overseas game expansion is also a major business focus of 37 Interactive Entertainment. In this area, "Xiaoqi" can also provide a lot of assistance. For example, in AI translation, more than 85% of the company's overseas games use AI translation, and the accuracy rate is as high as 95%. The AI material infringement detection function can quickly detect plagiarized advertising materials in the market, promptly notify the advertising placement personnel and legal personnel, optimize the efficiency of advertising placement, and safeguard the rights of the company against infringing materials.

These are the most obvious and commonly used improvements when AI is applied to the workflow. It can optimize many repetitive and meaningless operation processes, reduce costs, increase efficiency, and relieve people's work burdens.

There have long been many speculations on the Internet about AI replacing human jobs. Jokes like "Robots do the laundry and cook, while humans pursue poetry and the distant land" are everywhere. However, for some time, people found that human labor costs were much lower than those of robots, so no practical replacement solutions emerged. There was even the joke that "Humans continue to do the laundry and cook, while robots pursue poetry and the distant land."

However, the current trend of AI's practical application negates the view of "mutual replacement."

In the creative field, AI is more like a "helper" for creators. For example, with the help of AI agents, the art team of 37 Interactive Entertainment can produce more than 500,000 2D image materials per quarter, increasing efficiency by more than 80%. At the same time, they can also produce more than 300 3D assets, accounting for 30% of the total. In this way, the art team can devote more energy to creative design.

This is actually a consensus among many mainstream in - depth AI users. At this year's AI + Entertainment Technology Conference, Yang Sheng, the CEO of Polaris Interactive, said that "Technology serves people, and AI is a creative partner." Li Banqu, a well - known screenwriter, compared AI to a creator's "ideal co - pilot." Xia Lixue, the CEO of Wuwen Xinqiong, advocated that "AI changes the production mode." All of them emphasized this point.

After all, human values are what AI truly cannot replace.

02

Of course, finding the positioning is not difficult. The more crucial issue is "how to specifically use AI," which is also the core concern of most people.

37 Interactive Entertainment has gradually built a complete set of corresponding AI workflows.

"Xiaoqi" solves problems in internal office scenarios, such as generating meeting minutes, Word documents, PPTs, and posters. It also involves the combination of AI and production pipelines.

Before the popularity of AI, 37 Interactive Entertainment had already built a mature information and digital product system. For example, "Liangzi" is responsible for advertising placement, and "Tianji" is responsible for operation data. Now, as AI matures, intelligent technologies have been introduced into various products with clear divisions of labor, reshaping their cores and strengthening their functions.

Specific to each product, AI provides precision, high standards, and a corresponding set of big - data models. The specific problems they can solve are different. Therefore, employees must combine their needs to find the most ideal product and ultimately ensure the completion of content.

Of course, users don't need to consider which specific model is being used when using AI. This part is automatically scheduled and selected by the pre - set algorithms. The task of the AI R & D department is to ensure through pre - training and post - training that the models used are as targeted as possible.

Wang Chuanpeng emphasized that the ultimate goal is to "make users unaware of the use of AI."

He divided the application of AI into three stages, and "awareness" is the core criterion for judging the maturity of the field. The first stage is full - stack AI application, where all employees are aware of using AI. The second stage is like what 37 Interactive Entertainment is doing, integrating AI as an auxiliary partner and tool into the workflow. The third stage is AI - native application products, such as AI - driven games.

Currently, the first stage is like getting started with AI. People can choose AI tools according to different needs, and humans ultimately make key decisions. The second stage is that AI can completely subvert the previous production lines of products. For example, 37 Interactive Entertainment has built an AI architecture covering the intelligent product layer and the business application layer based on its AI capabilities, independently developed the industry - specific large - model "Xiaoqi" in the gaming field, and built the "Xiaoqi" agent platform, covering various business fields such as market analysis, product project establishment, art design, development assistance, advertising placement, operation analysis, intelligent customer service, and general office.

Full - link AI deployment of 37 Interactive Entertainment

The third stage is still in the exploratory stage. For example, Cai Haoyu launched "Whispers of the Stars," and Elon Musk has Ani. The limitations of these products are obvious. In some situations without clear settings, AI still cannot respond as flexibly as humans. This principle applies to any field, whether it's gaming, e - commerce, or social media. In short, AI still plays the role of a "productivity tool" and has yet to break through its own limitations.

03

It has to be admitted that on the one hand, AI can bring many disruptive changes to companies and projects, and even completely change the production mode and product line composition. On the other hand, AI in the present and near future still cannot do without humans, and humans remain the managers of production tools.

37 Interactive Entertainment

In response to this, Wang Chuanpeng put forward a very emotional perspective: "Let thinkers make choices, and let dreamers create."

He said that everything has its "ultimate value." For example, after AI subverts productivity, creators are more likely to make "choices" rather than "create." That is, after AI provides a large number of high - quality works, how should creators choose the products that better meet their needs?

In the past, creators could rely on their intuition to create and produce relatively ideal works, sometimes without even being able to explain why. After using AI, this situation no longer holds. For example, if a creator can choose a suitable AI work based on their "intuition," why can't a manager do the same? Therefore, AI works will force creators to think about why they make certain choices among many works. "They need to precipitate their experience to select the best works." This is not only about convincing the boss or the leader but also a process of convincing themselves.

The second half of the sentence is easier to understand. It is more targeted at the functions of AI itself. When using AI, developers can most intuitively feel what works well and what doesn't in AI tools, and they also have a clearer understanding of what an ideal AI tool should be like. Therefore, they can better help AI tools iterate and create more user - friendly products. This is also the consistent thinking of 37 Interactive Entertainment.

In essence, this sentence represents a process in which developers constantly "reflect on themselves."

Only when the level of creators is high enough can AI truly achieve some ideal results. Instead of becoming lazy after the emergence of AI, creators should strive for improvement. As Wang Chuanpeng also emphasized, for both 37 Interactive Entertainment and the entire industry, only when the level of industrialization is high enough can AI perform well. Relying entirely on AI in the future is not conducive to the development of AI itself.

Returning to the product itself, Wang Chuanpeng also constantly emphasizes that 37 Interactive Entertainment is still in the exploratory stage. No one dares to say that they have found the absolutely correct answer regarding AI. "We can only say that we have achieved some results, but there is still a long way to go in the future." Moreover, even before the emergence of AI, Wang Chuanpeng and his team had certain insights and ideas about informatization and digitalization, aiming to subvert the existing company structure and production mode through technology and look towards the long - term future.

As for AI itself, "the greatest certainty is that everything is uncertain." Because AI may have its limits, but human potential is infinite.

This article was first published on "36Kr Gaming".