Over the past week, I've delved into research on the gaming industry across various regions, as well as the burgeoning sectors of AI-driven short dramas and comic-based dramas. I'm grateful to friends who carved out time from their hectic schedules to dine or grab coffee with me, from whom I've gleaned invaluable insights and deepened my understanding of the current AIGC (AI-Generated Content) trend. My strongest takeaway is that the outside world seems to underestimate AI's impact on the traditional film and television industry while overestimating its transformative progress in the gaming sector. I won't delve into the former today, as that deserves a separate discussion; however, I have much to say about the latter.

Firstly, let's clarify—generative AI undoubtedly holds the potential to revolutionize the gaming industry, particularly in terms of cost reduction and efficiency enhancement. Yet, we should not overstate the significance of this transformation at its current stage:

1. AI programming tools, exemplified by Claude Code, and agent productivity tools, like Longxia, are indeed significantly boosting work efficiency within the gaming industry. However, this efficiency boost is not unique to gaming; it's applicable across almost all industries, with gaming not standing out as particularly exceptional.

2. AI-generated digital assets for games have made some strides, but their actual reach is not as extensive as the outside world perceives. The process of AI replacing game art and overhauling the entire technical pipeline will likely take more than 3-5 years to yield more substantial results.

3. As for integrating AI as a gaming experience, known as AI in Game, current attempts are primarily confined to independent and experimental works. Whether the 'freedom' and 'uncertainty' introduced by AI in Game can enhance the experience of mainstream players remains a highly debated topic.

Let's delve into the first two points. Both AI programming and AI-generated digital assets essentially aim to reduce costs and enhance efficiency. They significantly lower the barrier to entry for game development, enabling many small teams and even individuals to create their own games—a feat worthy of high recognition. They also, to some extent, lower the internal project approval threshold within large companies: now, producers in large firms can directly generate AIGC concept videos for new products based on core gameplay and art concepts to persuade executives to greenlight projects. However, it's fair to note that using concept videos (PVs) rather than demos as the basis for project approval predates the rise of AIGC; AI has merely streamlined this process further.

We can say that AI is reducing game development costs and facilitating more new products to reach the market—but so what? The gaming industry is inherently characterized by an extremely low success rate but high profitability upon success. Thousands of new games are launched globally across various channels each year, with over 2,000 games legally released in the domestic market alone. It's no exaggeration to say that in any gaming market or genre, supply far outstrips demand, with demand growth lagging significantly behind supply.

Now, AIGC further stimulates supply, which is certainly beneficial for changing the internal benefit distribution within the gaming industry, such as providing more opportunities to small companies and independent products. However, for the gaming industry as a whole, the allure of 'cost reduction and efficiency enhancement' is far from compelling. I recall that in a 2023 quarterly earnings call, the management of game engine company Unity stated, 'The revenue gap between successful and unsuccessful games can be tens or hundreds of times. Instead of trying to reduce development budgets by 10%, it's better to seriously improve the likelihood of success by a few percentage points.' Although this statement was essentially a defense of their business, the reasoning holds true.

I also want to point out that AI will not only increase the supply of so-called 'small-budget quality games' and 'independent games' but also flood the market with low-quality products, with the latter likely far outnumbering the former! The earliest large-scale adoption of AI-generated art assets was actually in casual games and small-to-medium-budget MMOs and SLGs—often derisively referred to as 'pay-to-win mobile games' by gaming purists. Many low-cost, IP-less, and short-lifecycle 'pay-to-win mobile games' have achieved even stricter cost control through AI, with the saved money naturally being funneled into user acquisition (don't assume they would invest it in content creation). In the long run, whether the internal benefit distribution within the gaming industry will evolve in a way that favors quality developers remains a significant question mark.

Frankly speaking, the gaming industry has witnessed rapid product evolution over the past decade. Almost all gameplay styles, genres, and business models have been thoroughly explored, resulting in a pervasive and persistent state of oversupply across the industry. Chinese game developers have certainly played a significant role in this, but European and American developers should not be underestimated—they are just slightly less 'aggressive' than their Chinese counterparts. The least 'aggressive' are probably traditional Japanese game giants, which is why Japanese games have been in a persistent decline in recent years, even being surpassed in traditional strong genres like anime-style games. The arrival of AIGC in this context is rather inopportune, as it will further exacerbate rather than alleviate the gaming industry's problems.

If we truly want to solve these problems, we must focus on the demand side—creating new demand. This is a topic that has often been envisioned since the emergence of ChatGPT: using AI to enhance game freedom, such as fully AI-izing NPCs in RPGs, AI-izing characters in dating simulation games, or even creating entirely AI-driven open worlds. In recent years, we have indeed seen some progress, with examples like the highly anticipated and media-covered 'History Simulator: Chongzhen' currently in testing. However, most major developers remain very cautious about this so-called AI in Game. Companies like Ubisoft and Square Enix have proposed AI NPC solutions but have yet to apply them to their flagship products.

Why is that? Let's pose a question: Are gamers truly pursuing freedom and uncertainty? If the answer is yes, then they shouldn't be playing games but rather 'playing' the real world, because in the real world, as long as you adhere to physical laws, you are free. You can stop a random passerby on the street, and their conversation with you will be completely uncertain; you can also try to do anything to them... as long as you can bear the consequences. If you're unwilling to bear severe consequences, try some harmless free activities, like kicking a roadside trash can, disrupting a group of aunties' square dancing, or interacting with cats and dogs. Isn't this freer than games? This is freedom that doesn't require any AIGC.

So why don't you like such freedom and uncertainty? Because you don't truly love freedom, and you especially hate uncertainty. When playing a dating game, you can only accept the outcome of successfully winning over your crush and developing a romantic relationship with them; when playing as Chongzhen, you can only accept the outcome of successfully saving the Ming Dynasty and 'Making Ming Great Again.' So-called uncertainty is just a slight fluctuation within this broad direction. The vast majority of players still seek the optimal solution within a certain range, which implies the best experience.

As a gamer with over thirty years of experience, my two favorite types of games are: first, Japanese-style RPGs, such as 'Expedition Team 33,' which I couldn't put down last year; and second, grand strategy SLGs, such as the 'Europa Universalis' series, in which I've spent thousands of hours. These two may seem very different, but their essence is the same, aligning with Sid Meier's definition of games: guiding players to make a series of meaningful decisions to achieve desired outcomes. The freedom I seek is the ability to enter scenes and defeat bosses in different orders on a large map or to complete tasks to become the greatest nation in the world according to different national policy trees. Such freedom existed long before the rise of AIGC. As for the freedoms that only AIGC can provide, I likely don't want them.

I especially don't want to find myself unable to judge the will of ministers and the people based on numerical values when playing as the Sultan of the Ottoman Empire or the King of England. If uncertainty doesn't affect core numerical values, such as only affecting narrative dialogue or producing a few beautiful narrative images, I can accept it; but I cannot accept ministers genuinely preventing me from conquering Constantinople, nor can I tolerate discovering that the event 'The City of World's Desire' has disappeared after conquering Constantinople because, according to AI simulations, this timeline has diverged, and Constantinople is no longer important... Such a game would be better off renamed 'Stellaris,' where all historical events are fictional. I'm curious to see whether AI-generated fictional events can resonate with me more.

Of course, perhaps in another five years, the vibrant development of the AI industry will render this article outdated—a prospect I would welcome, as my greatest wish as a longtime gamer is for the gaming industry to thrive. However, what I see now is that AI plays almost no substantive role in addressing gaming's 'demand problems'; while it can play a certain role in addressing 'supply problems,' the gaming industry is not particularly desperate for it. The capital market always wants to see AIGC transform traditional industries more quickly, but at least in the 'emerging traditional industry' of gaming, AI's transformative progress has been severely overestimated by the outside world. I believe that the more experienced the gamer, the more they will understand this point.