Insights into Web3 games: How AI creates a new "Yuan Shen"

Original author: Deon, GalaxyBlitz

Phenomenon Level《 Thoughts brought by "Yuan Shen"

Many people have found that facing "Yuan Shen", The traditional routines in the game circle have "failed". In the past mobile game industry, the "explosive models" born in the card, APRG, MMO, SLG, MOBA and eating chicken categories will allow the industry to quickly find the "direction", and many manufacturers will make a lot of money.

Facing "Yuan Shen", most people felt a huge sense of powerlessness. Most people want to copy homework, only to find that they only have a primary school education, and "Yuan Shen" is already an academician of the two academies. MiHoYo's open world, content packaging, large investment, industrialization, multi-platform, global distribution, IP and branding... have caused most people in the game circle to fall into a situation where they are unable to understand the answers.

Investors are more anxious. The industry giant only took 1 million yuan of investment from Song Tao, the CEO of Skye Network. Now the valuation of this investment is 340 billion yuan. The legend is comparable to Sun Zhengyi's investment in Alibaba. So where is the next Mihayou?

The AI Being full of personality will bring the narrative climax of the next Web3 game.

Let's first observe the original God's payment logic, the character is the core payment point of "Original God". "Original God" is a game whose content is extremely centered around the character. It is to serve the role. Each character will have its own plot and story, and even the improved game experience of the individually produced big world (such as the layered rock abyss) will be shared with each character, because you still have to use them to explore in the end.

So why do players pay for characters? Because the players have feelings for the characters, the Tivat continent is indeed a beautiful stage. The big world created by MiHoYo, the accompanying reverberating music, and various characters and stories all make people deeply attracted. Walking on this land is like walking in a real different world - not limited to novels, animations, and truly feel an idyllic poem from a strange land. Beautiful things are naturally alluring, and we flock to them.

In this beautiful continent, whether players and characters are exploring, fighting, clearing the abyss, or cultivating, To some extent, it makes you think that you are really in the wilderness or raising children, and your brain will produce a certain pleasant feeling. This pleasant feeling comes from the deep relationship with the player and the two-dimensional characters in the next adventure. Impressed by the character experience and personality.

For example, Walnut is a classic "contradictory" character design, both lively and deep, both understand The dogma of life and death is free and easy. On the surface, he doesn't care about anything, but in his heart he is very serious about death. This kind of complex and contradictory three-dimensional character, on the contrary, makes people feel the charm of the character of Walnut even more.

The surface of the walnut is naturally playful , free and easy, but after a deep understanding of her experience, I know that she has a solemn and dignified side

Mihayou gives each role Created a highly three-dimensional and complex multifaceted character. After experiencing and exploring together, players gradually understand and dig out the story of each character. They are impressed by the character and charm of the character, and have the desire to collect it, so they have the urge to pay.

This is also the reason why "Original God" has to undergo high-intensity updates continuously, because only a large number of high-quality content updates centered around the characters can attract players and characters to have empathy and a sense of intimacy in the adventure, thus Attract players into the pay loop.

Web3 field SLG will use AI technology to create a new "Yuan Shen"

So can such payment logic be reproduced in the SLG type? Obviously, the development of AI technology has made this logic possible. GalaxyBlitz will use AI technology in the next version, so that every character in GalaxyBlitz becomes an AI Being that can absorb the character of the protagonist. Unsupervised reinforcement learning technology can make AI Being similar to human cognitive awareness. All actions of the character, such as swinging a sword, Dodging, defending, walking, sprinting, bending, shooting, etc., are all produced by the player's own training, with the imprint of the protagonist, making decisions similar to the player's, and each character essentially grows with the player , AI relatives who have experienced together, are labeled as players, and it is very easy to empathize with players.

How to implement these AI technologies? GalaxyBlitz and Phantom AI, a well-known AI company, have developed GalaxyBlitz Mobile Net, a neural network training system suitable for SLG mobile games. GalaxyBlitz Mobile Net is a scalable data-driven method that uses low-power player mobile devices and GalaxyBlitz's deep data enhancement. Together with the center, learn AI Being with reusable motor skills.

Our framework is outlined below:

GalaxyBlitz Mobile Net Framework

GalaxyBlitz Mobile Net framework consists of two stages: pre-training, target training.

During pre-training: low-level policy (a 丨 s, z) is defined as using a as the basic action in the action space data set s as a variable Implement a map of goals z that model various actions with a reward function r. After pre-training, it can be transferred to the target task by using a specific task.

low-level policy can model various general actions, according to player habits defined by Environment0 The state set can realize the training of general actions such as swinging a sword, walking, sprinting, bending, and shooting.

Low-level policy action combination - wading< /blockquote>

Low-level policy action combination - roll and jump

Low-level policy action combination - gun posture running

Low-level policy action combination - kneeling shooting

During target training: High-level policy w (s 丨 z, g) is defined as one of the action space data set s used as the action combination z as a variable to achieve the target g map, where z = array(a).

As shown in the figure below, through the High-level policy, the AI of the Low-level policy has been learned Being can be instructed to perform various specific tasks, which are composed of a set of actions in the Low-level policy.

High-level policy Combination of Action Tasks - Kneeling and Shooting the Bullseye

High-level policy action task Combination - walk around the flag

In the High-level policy, the random sample generation of Environment1 starts With the combination of simulated character performance skills, as the player operates frequently, the discriminator will separate the actions that the player is accustomed to, and the actions of AI Being will become more and more in line with the player's habits.

As shown in the figure below, our actions do not include "getting down" and "standing up", but AI Being will learn "getting down" and "standing up" on its own stand up". If the player likes the timid character of AI Being, then AI Being will "fall down" when encountering danger, and "stand up bravely" at critical moments, so as to discover the shining point in the player's heart.

High-level policy Let AI Being has the player's character of "usually timid but courageous at critical moments"

The rise of the new Web3.0 SLG< /h3>

If the current version of GalaxyBlitz 1.0 finally achieves an annual turnover of 200 million US dollars, it will reach a relatively high level of SLG in the current Web2 field. But GalaxyBlitz will be realized in the next AI version: investment demand + entertainment demand + emotional consumption demand, which is expected to reach an annual turnover of US$5 billion similar to the level of "Yuan Shen".

Gamefi as envisioned by the GalaxyBlitz team Evolution route

The new SLG represented by GalaxyBlitz AI technology will have thousands of faces Character NFT, in the next generation of SLG, each character has the player's unique preferences and data, they are no longer the stereotyped numerical tools in traditional games, but rely on the player's emotions, playability, scalability, etc. be greatly improved.

< p>Therefore, AI Being full of personality will bring the climax of the next Web3 game narrative.

This article is from a contribution and does not represent the views of BlockBeats

Original link

Welcome to join the official BlockBeats community:

Telegram Subscription Group: https://t.me/theblockbeats

Telegram Discussion Group: https://t.me/BlockBeats_App

Official Twitter Account: https://twitter.com/BlockBeatsAsia