IbizaPocholo
NeoGAFs Kent Brockman
Machine learning has historically not been seen as a practical solution for many of the challenges that video game developers face, especially in non-player character design, but recent innovations in the field, particularly the boom of research in deep learning, are beginning to change that.- Machine learning is increasingly practical to apply in the games industry, and it is beginning to fundamentally change how video games are being made.
- Machine learning is used in player modeling, analytics, cheat detection, texture synthesis, graphical upscaling, animation blending, and much more.
A machine learning system is built "to adapt to new circumstances and to detect and extrapolate patterns", and as these systems adapt and extrapolate these patterns, they improve at the task they're set, which is observed as it learning.- The reasons for using a machine learning system are: designers cannot anticipate all possible circumstances that the agent might find itself in, designers cannot anticipate all changes over time, and sometimes human programmers have no idea how to program a solution themselves.
- Machine learning is a conceptually old as other forms of AI and is not a new concept.
- Machine learning was a dead end and failed to work at scale or with larger, more complex problems while symbolic AI continued to deliver. However, machine learning has seen a new lease on life over the past decade, courtesy of breakthroughs in research and large-scale cloud computation platforms.
- Machine Learning wasn't practical for games before, and its use in games was limited to some notable examples such as Forza's Drivatar, Lionhead's Black & White, and Creatures.
- With recent advancements in AI research, more games are using machine learning for NPCs and opponents, ranging from the AI drivers in the MotoGP series to the enemy AI in Hello Neighbor 2, and the campaign AI in Age of Empires IV.
The real AI revolution happening in the games industry is occurring in other areas of video game production, such as motion matching, texture upscaling, upscaling graphics in real-time, and quality assurance.
AI systems can playtest games and help alleviate a lot of the pressure being placed on QA teams. Companies like modl.ai are developing tools that run in-game engines to facilitate this for smaller dev teams.
EA uses bots trained with machine learning to test games and find issues, starting with Battlefield V back in 2018 and continuing to update their testing tools for various games such as Star Wars Battlefront 2, Apex Legends, and Battlefield 2042.- Machine learning is being used for cheat detection, recognizing and identifying erroneous behavior by players, which is often challenging.
Last edited:
