Notes

Minecraft Diamond Challenge Leaves AI Creators Stumped

Minecraft diamond challenge leaves AI creators confused By Sam Shead Technology reporter



12 December 2019



It takes minutes for most new Minecraft players to work out how to mine the diamonds that are key to the game, but training artificial intelligence to do this has been more difficult than anticipated.



In the summer, Minecraft publisher Microsoft and other organisations challenged coders to develop AI agents that could find the coveted gems.



Most people can master it in their first session.



However, of the more than 660 entries, not one of them was up to the task.



The official announcement of the results of the MineRL competition which is known as "mineral", will be announced at the NeurIPSAI conference in Vancouver, Canada on Saturday.



The objective was to determine if the problem could be solved using minimal computing power.



Despite the absence of any winners, one of the organisers stated that she was "hugely amazed" by a few participants.



"The task we asked for is very hard," said Katja Hofmann, a principal researcher at Microsoft Research. "Finding a Minecraft diamond is a process that involves many steps, from cutting down trees to making tools , to exploring caves to actually finding a gem.



"While no agent submitted has yet solved the problem completely, they have made a significant amount of progress and have learned to make many of the tools needed in the process."



Mining diamonds



Minecraft has grown in popularity since its release in 2011.



More than 180 million copies of the game that is open-world have been sold, and the game has more than 112 million active monthly players.



Diamond is one of the most important resources in Minecraft because it can be used to make strong armour and powerful weapons.



However, in order to acquire the precious stone, a player must first complete a number of other steps.



"If you're familiar with the game it shouldn't take you more than 20 minutes to earn your first diamonds," Minecraft player Jules Portelly told the BBC.



Entrants were only allowed to utilize a single graphics processor unit (GPU) and four days of training time. AI systems often require months or even years of gaming time to master games like StarCraft II.



To train their systems, a relatively small Minecraft dataset with 60 million frames of player data was made available to all participants.



"At the start of every episode they spawned in a procedurally-generated Minecraft world," explained Dr Hofmann.



"So they really had to be able to comprehend the concept of finding resources, making tools , and finding the diamond."



Minecraft to run artificial intelligence experiments



DeepMind AI achieves Grandmaster status at Starcraft 2



Minecraft players will be assisted by AI assistant



The organizers wanted the programmers to develop programs that learn by example, through an approach called "imitation learning".



This is where you try to make AI agents to take the best approach to solving a task by getting them to imitate humans or other software.



It differs from relying only on "reinforcement learning", in which an agent is effectively taught to find the best solution via a process of trial and error, but without drawing on past knowledge.
35eng


All AI



Researchers have found that reinforcement learning on its own can produce superior results.



For instance DeepMind's AlphaGo Zero program trumped one of the research hub's earlier efforts, which used both reinforcement learning and the study of data labelled from human play to learn the board game Go.



This "pure" approach requires more computing power and is too expensive for researchers who are not employed in large organizations or governments.



William Guss, a Carnegie Mellon University PhD student, was the principal organizer of the competition. He stated that the purpose of the contest was to prove that "throwing massive computation at problems isn’t necessarily the right way to advance the state of the art as a field."



He said, "It works against democratising access to these systems of reinforcement learning and leaves the capacity for agents to learn in complex environments to corporations that have vast areas of computing."



However, the outcome may reveal the advantages these entities with a good financial position have.



Minecraft graphics to look more real



19 August 2019



Minecraft player loses five-year-long game



29 April 2019


Website: https://35eng.com/