Site icon The East County Gazette

Facebook wants an AI that can beat the ‘hardest game in the world’!

facebook NetHack

Today’s social media giant Facebook proposed an idea to beat the world’s most challenging game – NetHack. The company proposed this idea NetHack for a grand challenge using AI, and the company plans to launch this competition at the NeurIPS 2021 AI conference in Sydney, Australia. 

NetHack, an 80’s game with simple graphics and interface, is the world’s most brutal game to win and battle. AI can help scientists create algorithms using technology that can withstand the most challenging times without running experiments on a powerful computer. Real-time strategy, role-playing, and building games help train cutting-edge technology that uses artificial intelligence. 

Facebook wants to challenge the game and maybe wants to help computers learn to simulate situations faster using minimal resources. Games have always been AI benchmark for AI for several generations, but things got spicier in 2013. 

In 2013 Google DeepMind illustrated a system that can play different games like Pong, Breakout, Space Invaders, Seaquest, Beamrider, Enduro at superhuman levels which were only imagined. The technology is improving the game design, according to the review by an expert at DeepMind cofounder Demis Hassabis. But these technological advancements could be for  generations one day to diagnose and treat illness, predict complicated protein structures and CT scans. 

Facebook is inviting different researchers worldwide to design, train and release AI systems as part of the NeurIPS 2021 NetHack Challenge. According to the blog AI can, “develop agents that can reliably either beat the game or (in the more likely scenario) achieve as high a score as possible.”

Facebook plans to showcase the NetHack learning environment and see the potential of new and potential AI/ML technologies based on both neural and illustrative methods.

“The candidate agents will play a number of games, each with a randomly drawn character role and fantasy race,” the Wednesday post explained.

 “For a given set of evaluation episodes for an agent, the average number of episodes where the agent completes the game will be computed, along with the median in-game end-of-episode score. Entries rankings will be by average number of wins and, if tied, by median score.”

The contest will start on October 15, and the winners will be announced at the end of the year. 

Exit mobile version