#AI—Arago, a pioneer in artificial intelligence (AI) and leader in intelligent IT automation based in Frankfurt and New York City, today announced that HIRO (Human Intelligence Robotically Optimized)™, the company’s flagship AI technology capable of learning, understanding and reasoning to solve problems, has defeated human players in the computer game Freeciv, after it became an unbeatable opponent for the game’s built-in AI. Freeciv is a turn-based strategy game inspired by the popular Sid Meier’s Civilization series. This signifies a major step forward in one of the challenges in the development of AI: evolving decisions from brute force testing, to more nuanced role playing without defined outcomes for victory.
The announcement, which included game footage screened Tuesday afternoon at TechCrunch’s Disrupt London 2016, represents an exponential advancement in machine intelligence due to Freeciv’s game complexity. With all the different options, opponents acting unpredictably, and changing parameters throughout the whole game, there are more than 10 to the power of 15,000 possible games imaginable – for comparison there are only 10 to the power of 80 atoms in the universe. Advancements from simple trial and error represented by machine learning, to combining this approach with machine reasoning that takes the “why” into account was Arago’s goal.
“It has become good tradition to showcase the advancement in AI through playing games; starting at chess to Jeopardy to Atari games to Go. Winning in a strategy game like Freeciv is the next logical step,” said Arago founder and CEO Chris Boos.“Beyond the simple growth of complexity, Freeciv introduces additional layers of realism by adding
chance and no longer having god-like information about the game. HIRO required less time and resources for training than previous AI showcases. We’re able to demonstrate the power of the hybrid platform based on machine reasoning enhanced with machine learning underlying the HIRO system.“
On top of the scientific showcase of winning Freeciv the same HIRO system is used by global enterprises for automating IT operations across the entire IT stack. Commercially, the HIRO platform amplifies human IT experts through continuous learning and built-in self-optimization.
Faster Learning through Optimized Efficiency
Despite the fact that Freeciv holds an exponential amount more moves than Go, Arago’s HIRO actually required less data and trainers in order to advance to victory. Winning against human players is a major milestone, because it shows that hybrid systems of machine reasoning and machine learning can reuse their knowledge and though easily be trained by different independent coaches simultaneously.
“A common misconception with regards to AI can stem from popular culture and fantasy, that we are looking for cognizance or singularity,” added Boos. “In fact, we are looking for machines with memory and adaptation for long-term planning. These properties give HIRO the capabilities to excel in very complex situations such as IT system management and also in Freeciv.”
Boos spoke with former TechCrunch Editor John Biggs onstage at TechCrunch’s annual Disrupt conference in London where the crowd was treated to a time-lapsed encapsulation of wins against humans and other outstanding scenes from game play.