Notes

Minecraft Diamond Challenge Leaves AI Creators Stuck

Minecraft diamond challenge leaves AI creators baffled by Sam Shead Technology reporter



12 December 2019



Most Minecraft players can find the key diamonds in minutes. However, it is much more difficult to create artificial intelligence that can do this task.



Microsoft, an Minecraft publisher, and other organizations challenged coders to develop AI agents that could locate the sought-after gems.



The majority of people can complete it in their first session.



However there was no way to prove that any of the more than 660 entries submitted met the requirements.



The results of the MineRL - which is pronounced mineral-related competition is scheduled to be announced officially on Saturday at the NeurIPS AI conference in Vancouver, Canada.



The aim had been to determine whether the issue could be solved without massive amounts of computing power.



Despite the absence of an award, one of the organisers stated that she was "hugely impressed" by a few participants.



Katja Hofmann, the principal researcher at Microsoft Research, stated that "the task we set was very difficult." "Finding diamonds in Minecraft requires a variety of steps ranging from cutting trees, to making tools and exploring caves, before actually finding a diamond.



"While the submitted agents aren't able to solve the task fully however, they have made significant progress and have learned how to use many of the tools needed."



Mining diamonds



Minecraft has grown in popularity since it was released in 2011.



The open-world game has been downloaded more than 180 million times, and the game is played by more than 112 million people each month.



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



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



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


The players were allowed to use a single graphics processing unit (GPU) and four days of time to train. For context, AI systems usually need months or years of game play to master games such as StarCraft II.



A relatively small Minecraft dataset, comprising 60 million frames of human player data, was also made available to participants to train their systems.



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



"So they really needed to learn the idea of locating resources, making tools and finding a diamond."



Minecraft to run artificial intelligence tests



DeepMind AI attains Grandmaster status at Starcraft 2



AI assistant to assist Minecraft players



The organisers wanted the coders to develop programs that could learn by example, through an approach called "imitation learning".



This is where you try to get AI agents adopt the best approach to solve a problem by making them imitate humans or other software.



This contrasts with "reinforcement learning", where an agent is trained to find the best solution by trial and error.



AI for all



Researchers have found that using reinforcement learning alone can sometimes deliver superior results.



DeepMind's AlphaGo Zero program, for instance, beat an earlier effort by the research center that utilized reinforcement learning and the analysis of data labelled from human behaviour to learn the board game Go.



However, this "pure" approach typically requires more computing power, making it too expensive for researchers other than large organisations or governments.



William Guss, the main competition organiser and a PhD student at Carnegie Mellon University, told the BBC that the purpose of the contest was to demonstrate that "throwing massive computation at problems isn't necessarily the best way to push the state of the art in a field".



He added: "It works directly against democratising access to these systems for reinforcement learning, and leaves the ability to train agents in difficult environments to companies with vast amounts of compute."



However, the final outcome could be a good indicator of the advantages these entities with a good financial position have.



Minecraft graphics to look more real



19 August 2019



Minecraft player is sacked for five years



29 April 2019


Homepage: https://faxsys.com/