Google DeepMind AI Beats Human Players in Quake III: What Comes Next?

Training artificial intelligence through multiplayer gaming is a niche which is, believe it or not, a pet project of Google’s DeepMind AI lab. As reported in Science recently, the project has successfully ‘learned’ to play Quake III. After feeding the program thousands of hours of footage and game data, it was able to take on humans with great success. The maps were randomly generated but that did not stop the AI from crushing its human competition. On average, the AI was able to beat humans by a margin of 16 captures.

Versus AI: Two Heads Are Better Than One

Humans were only successful against Google’s AI when they teamed up together. For example, a team consisting of one human and one AI had a five percent greater chance of winning than a team of just AI. Emerging technologies like AI are quickly improving, but game-based learning is unique. It requires looking at hundreds of thousands of possible scenarios — a good vacuum for testing AI. DeepMind AI still can’t take on the best Halo players, but it is coming close.

Google is Leveraging Emerging Technologies

Google’s DeepMind has been doing a whole lot more than just teaching AI to play video games. Back in 2017, the Google subsidiary rolled out a major project with the NHS to bring patient information on a distributed ledger system. It is clear that Google sees blockchain technology and AI as closely related by using machine learning to inform decentralized systems. It is a concept that has a longstanding history in the existing cryptocurrency market — with various projects existing which tackle questions of artificial intelligence. Arguably, we need both distributed ledger systems and AI to create organic networks which promote machine learning. Google’s DeepMind recognizes this fact. Yet, DeepMind just may focus on fine-tuning its AI’s gaming capabilities for the time being. After all, it’s probably fun to watch the AI you created frag even the best Quake III players. Do you agree that AI and decentralized go hand-in-hand? Let us know your thoughts in the comments below.