Artificial intelligence, the new video game revolution?

Artificial intelligence, the new video game revolution?

The AI field may have been one of the most abandoned in the history of video games. Since Pacman, the way to program behavior routines has not changed excessively. However, it seems that there is a great transformation, and it has to do with two terms: autonomous learning and data.

we will be more progress in the next five years than we have seen in the last 45″. These were the overwhelming words of Patrick Soderlund (former EA executive) when speaking not of streaming game, or ray tracing technology … but of something that is not making so much noise, but that has been investigated for years.It is powerful, and a good number of companies are working on it, with million-dollar investments. Artificial intelligence can become the next great revolution of the video game industry .

The investment in general AI by companies has grown to 5,000 million dollars in less than a decade, with fierce progress in recent years. Google is posed as the master and lord in this field, not only in relation to its mythical apps, such as Google Maps, but also in video games. For more details, we have been seeing how Electronic Arts, Ubisoft and even Square Enix are taking the issue into account. Significant industry voices, such as veteran developer Tim Willits (id Software), left no doubt, ensuring that the next generation’s great contribution will be artificial intelligence.

It is not an AI programmed exclusively by developers, but the concept goes further. The future is considered as artificial neural networks that learn from the successes and failures of the CPU, as well as from the observation of real players. This intelligence, successfully tested in video games like Quake III or Starcraft II, is not a mere simulation or prototype. It is already a complete reality.

It is an advance difficult to understand, with many edges to deal with, and still has several challenges to overcome. Our goal here is to explain to you who are the main actors (or at least those who have spoken publicly on the subject), what is being done, and why it is important to understand what we are going to see in the next few years.


If you see Stadia’s presentation again, you will notice something very important: the first thing the executive director Sundar Pichai spoke about was artificial intelligence. But not just anyone, but one that “can learn to solve problems.” He even spoke that these advances, experienced in the world of videogames, serve the company to put them into practice in its applications for the real world.

What is AI proposal?
The key to understanding what Google is doing has a name: DeepMind. It is a British subsidiary that has managed to create a neural network capable of learning from the actions of the players. First he did it through the AlphaGo computer program (2015), defeating a professional player of the strategy game Go. His move to video games was recently made with AlphaStar (2019), by defeating several professional Starcraft II players.

Operation and applications
DeepMind AI technology is based on reinforced learning. For practical purposes, this involves creating a CPU-driven player that can learn with each game and solve problems without any instruction. How?

DeepMind reached an agreement with Blizzard with which to study a large number of user games in the Starcraft II strategy video game. Specifically, AlphaStar could learn in one week the equivalent of 200 years of games.

The results are spectacular. The program managed to defeat professional players in ten games in a row, always adopting different strategies, and without repeating any. The most curious thing is that the program does not perform more actions than its rivals, but more effective.


Although it doesn’t make much noise, the gala company has been working on different artificial intelligence solutions for some time. Ubisoft La Forge is a prototyping space that shares the advances made in the academic world (university researchers) with the company. In this regard, it is interesting to note that the advances that are being made are not only directed at the games themselves, but also at the way they are developed, optimizing work resources.

What is AI proposal?
One of Ubisoft’s most promising applications is SmartBots, which would solve one of the greatest difficulties that open worlds have: their increasing complexity. This application is designed to have increasingly real NPCs, whether it is an allied AI as an enemy. The system observes the actions of these bots and establishes a point system, so that successful actions are rewarded. In other words, an artificial intelligence is generated that learns with each game.

Operation and applications
Ubisoft promises that the new generation bots will be credible characters capable of driving vehicles and accompanying you or fighting you. However, another of the most interesting applications of AI has to do with the development of a wizard that helps predict and eliminate bugs. It ensures that developers can save 20% of their work time and that information about possible bugs arrives even before they occur. Basically, it works through a database of lines of code collected for 10 years, which help to compare structures and thus check what can and can not work.


The investment in the new advances of AI is also found in the traditional third-parties of the videogame industry. EA believes that artificial intelligence will have a considerable impact within three to five years. They tell from the North American company that when they saw the technology of DeepMind in 2015 they were speechless. Since then, they focus their efforts on applying these advances to triple A games.

What is AI proposal?
Electronic Arts has prepared and created SEED (Research Division of Extraordinary Experiences). His approach is very similar to that of DeepMind with its reinforced learning logic, only in this case applied to large productions of the Battlefield style. In this sense, they intend to replace the classic bots by what they are known as “self-directed agents”, who expect them to be able to defeat – within small maps and with defined objectives – real players within two years.

Operation and applications
The artificial intelligence proposed by EA has already been applied to Battlefield 1, resulting in a program that learns from its mistakes. AI reinvents its way of acting, so that it changes its behavior if it has poor health or ammunition. From the company they are sincere saying that, at the moment, their program is not as skilled as human players, and that they behave erratically in certain situations. However, they establish that the difference is not very large, and they are hopeful that this will contribute significantly to the development of video games. As a detail, Magnus Nordin, technical director, said that in these tests the real players did not know how to identify the characters managed by CPU, since their behavior was quite real.


The president of the company, Yosuke Masuda, defends that the video game industry will undergo various changes, and one of the main ones will come from artificial intelligence. The leader in the AI department of Square Enix, Youichiro Miyake, has come to say that it will be “the next revolution”, something that machines like PS5 or the next Xbox will make possible. The Japanese giant seems to be especially prepared for what is coming.

What is AI proposal?
Youichiro Miyake is a researcher at Square Enix who is convinced that we are living the third boom in terms of artificial intelligence, this time based on terms such as “deep learning” or “big data”. That is, systems that learn by accumulating large amounts of data, which go to an artificial network that interprets them. Until now, AI was based solely on rules set by programmers, but Miyake believes that the future is about seeing machines that react to the style of play, and even the personality of the players. He does not believe that the figure of the programmer disappears, but it will be complemented by a concept of adaptive artificial intelligence, which responds dynamically to the actions of the players.

Operation and applications
The developers of Square Enix say they have learned a lot from the development of Final Fantasy XV. They are currently studying how to integrate a development structure that is not only based on human programming, but also on the programming that the machine automatically carries out. Here they talk about content generated by AI (map), self-adjustment of difficulty by AI, search for bugs by AI … and the one known as AI of crowd, not based on intelligent objects, but in areas that developers program to present different behaviors Hajime Tabata, involved in the development of FFXV, went on to say that he believes that the next project (a new Final Fantasy?) Will be something that only they can achieve based on the knowledge acquired about AI.


A new generation of AI
All exposed companies (Google, EA, Ubisoft and Square Enix) coincide in an artificial intelligence that learns, handling large amounts of data. As Tim Willits (id Software) says, it is “an AI that works as a game master, capable of making an experience as a human designer would.” The applications that interest us most are in the field of gameplay design. There have already been cases like bots that can play Quake III better than us, learning with each new game. Other applications are also being evaluated, such as detecting toxic players, eliminating bugs, or even performing lip sync by AI.

The technological base
The progress of AI that is being developed makes use of a large amount of data, and takes into account multiple variables. It is logical that the better the technological base, the greater the possibilities. Nvidia has been focusing in recent years on the development of AI systems, in what they call “the ultimate gaming experience.” For its part, the manufacturer AMD has been working on a 7-nanometer graphics chip optimized for artificial intelligence work. Of course, in the world of home video consoles, it is expected that both PS5 and Xbox Scorpio will be prepared to promote this jump, something that in Stadia would already be planned thanks to its scalable hardware.

When will we see these advances in AI?
Based on the information presented, we will see them sooner than we expect. In fact, we already have examples. Julien Merceron (head of the Fox Engine), says the Forza series relies on this technology. Also, Patrick Söderlund (former head of Battlefield) states that the industry is going in this direction, and that we will see it “in a short time”.

Leave a Comment