Huang Renxun introduces Nvidia G-Assist, which appears to be just an AI game assistant, but the underlying principles may disrupt the gaming industry, leading to the accelerated elimination of 3A factories and pay-to-win games, and sparking a revolution in the gaming industry.
Next decade, would you bet on Nvidia or Bitcoin?
Background:
Nvidia has launched a new AI game language model called Project G-Assist. How will it change the GameFi track?
Table of Contents:
AI Moves from Visuals to Game Play
AI Will Eliminate Unfun Games
Blockchain Games: P2E Will Disappear
Huang Renxun came to Taiwan to deliver a speech at COMPUTEX 2024, announcing the revolutionary AI game assistant, GeForce Project G-Assist. Although it seems to be only a game assistant that helps players strategize, analyzing its underlying principles and applications reveals that Nvidia, a leading AI hardware manufacturer, has opened up a new direction for the global gaming industry for the first time. This will have a profound impact on the development and business of the gaming industry. This article attempts to analyze in depth the future trend of the combination of the gaming industry and AI, as well as the trends in the GameFi track of blockchain games.
Nvidia started with game graphics cards and drivers, and in April 2020, they made a groundbreaking integration of AI and games by introducing Deep Learning Super Sampling (DLSS) for use in the RTX series graphics cards, ushering in a new era of gaming.
DLSS works by using a large number of Nvidia game graphics cards and backend game calculations to analyze and calculate inefficient operations that are imperceptible to the naked eye at high resolutions. Through AI models, it “reduces performance load” on graphics, resulting in revolutionary functionality with no compromise on visuals but a significant increase in frames per second (FPS).
After the outbreak of the LLM language model in 2023, OpenAI initiated a computational war using Nvidia graphics cards, with many software giants, including Microsoft, Google, and Musk’s XAI, participating. Although Nvidia, as a graphics card manufacturer, seems relaxed, they have not forgotten to innovate in software. Today, they have launched Project G-Assist, solidifying Nvidia’s ambition to integrate the LLM language model with games. This is bound to lead a new wave of revolution.
According to Nvidia, G-Assist can provide gameplay recommendations and story guidance for games. Players can ask the robot for equipment recommendations. In fact, it collects a large amount of gameplay data from players and feeds it into a large-scale language model for learning. This can effectively lower the threshold for players to play games and create more consistent game services. This will fundamentally change game production and gameplay ecology.
How will G-Assist change the gaming ecology? Imagine if games, from alpha and beta to release, are all collected by large language models based on game levels. What kind of situation will arise? The result is that games that lack diversity and have poor gameplay will be easily found by machine learning to have the best solution. The gap between professional esports players and regular beginners in immature games will be minimized, or it will bring the following impacts:
Minimization of beginner tutorials
Template-based skin-changing games become boring more quickly (applies to both mobile and 3A games)
Failure to create diverse games will fail to attract excellent players
AI will immediately expose the pay-to-win trap
Accelerated revision and iteration of esports games
Currently, the gaming industry is filled with a large number of mobile games and 3A games that have similar gameplay experiences. Although these games claim to be diverse, they actually follow a standardized production process, resulting in limited gameplay and variation across different games.
In reality, the experience between different gameplay styles, influenced by individual differences in gameplay (such as watching walkthroughs and pay-to-win), actually determines different gameplay experiences. If these differences in gameplay experiences can be quickly optimized, for example, taking the skin of Game A and applying it to Game B, the model generated by Game A will instantly train Game B to the best solution.
As a result, the current practice of large game companies using a template to produce multiple games will rapidly digitize gameplay. Therefore, 3A games that only change skins and mobile games that only change skins will inevitably face challenges.
Taking mobile games as an example, imagine if AI has already played a large number of poorly designed pay-to-win mobile games. AI can quickly calculate how long it will take for you to encounter frustration if you don’t pay, how much time it will take to reach a certain level by playing for free, and how big the gap is between paying players and you. This will instantly expose the “psychology of pay-to-win” meticulously designed by the developers. If AI is not restricted by developers, players will be able to see through everything through AI from the beginning.
Therefore, in the future, game developers will inevitably be divided into two factions. The first is the faction that caters to AI, utilizing AI data learning, diversification, and balancing various gameplay experiences. The other faction is the faction that restricts AI, using APIs and encryption to limit G-Assist or other open-source game language models, forming a new battle of data interpretation to protect commercial interests.
Although it seems unrelated to the blockchain industry and GameFi, web3 projects and communities have embraced the spirit of open source. If the code and immature games or Ponzi schemes are not well-developed, they will undoubtedly be quickly learned by G-Assist and similar language model competitors, and it will be immediately apparent that these games have “no future.” Therefore, under the popularity of G-Assist and similar models, short-lived games and scams that lack sustainability in “Play To Earn,” VC-driven valuations, Ponzi schemes may be directly terminated by technology.
On the other hand, true blockchain games with gameplay and diverse mechanics that can withstand AI testing may have a high chance of surviving. This is both a joy and a concern for the web3 game industry. This means that no matter how much funding they have or how luxurious their games may seem, their gameplay will be exposed under technology. This is undoubtedly good news for players, but it also limits the marketing methods of future game industry professionals.
Regardless, with the introduction of G-Assist and similar technologies, the gaming industry is bound to experience a fierce battle. Formulaic high-cost productions of 3A games may be significantly challenged, while low-cost, creatively replayable independent games may stand out. However, this may only be possible after several iterations of new model technologies such as G-Assist.
Related Reports
Nvidia’s stock price surpasses $1100, reaching a new high! Huang Renxun praises: Taiwan is the center of global AI.
What AI tools does Huang Renxun use? Perplexity AI ranks among the best, what about other tech giants?
Meta’s Chief AI Scientist, Yang Likun, strongly criticizes Musk: spreading crazy conspiracy theories, a selfish and self-serving hype king.