Minecraft
game
OpenAI
video
contractor
model
create diamond tools
Minecraft
game
OpenAI
video
contractor
model
create diamond tools
Ai Learned to Play Minecraft and it's actually Pretty Good Chess, Go, and now-Minecraft. Artificial intelligence models have added a brand new victory to their killing list. - Off - English Artificial intelligence company OpenAI was able, using an assortment of unlabeled Minecraft videos, as well as the small percentage of those that were labeled by contractors, to build an artificial neural network that could play Minecraft. This is a significant achievement for the technology which previously had a difficult time overcoming the game's simple but loose gameplay. Open AI engineers published their experiment in an article and blog post this week. OpenAI's model could be more than just basic survival and crafting. It could perform many of the same complex tasks as a human Minecraft player. OpenAI released a video that shows its model swimming, hunting and cooking animals. It even successfully figured out the game's "pillar jumping" technique. In the past, Deepmind was able to successfully make its MuZero AI play Atari games. In order to beat classic games like Chess and Go, previous AI models relied on reinforcement learning. Minecraft On the other hand is a game that can be understood by children of all ages however, it presents challenges for AI systems due its open-world design and open-ended structure. While there's a seemingly endless number of videos on the internet about Minecraft gameplay, those only tell a portion of the story of how to learn to play the game, at a minimum when training an AI. OpenAI claims that the unlabeled video data does a good job showing "what" but it doesn't provide precise keystrokes or mouse clicks that are required for an AI understanding "how" to play. The engineers solved this "how" problem by creating an imitation learning technique that is semi-supervised they call "Video PreTraining," or VPT. OpenAI basically gathered a smaller and more compact dataset from contractors which included not just Minecraft gameplay but also examples of key presses and other actions documented by the contractors. OpenAI created another model that relies on videos from contractors to predict the next action in each step of a Minecraft film. Their AI was able to comprehend large quantities of Minecraft videos online once it had the basic. Instead of simply dump a large amount of data onto their AIs, the engineers took the time to teach them the fundamentals of inputs. Minecraft "For many tasks our models show human-like performance and we are the first to report computer agents that create diamond tools that can take proficient humans upwards of 20 minutes (24,000 actions in the environment) of gaming to complete," OpenAI worte in their research paper detailing the results. According ZDNet, the price of all this training and contractor assistance was around $160,000. According to ZDNet the majority of the money was paid to contractors who made up around 4,500 hours of gameplay. The contractors were paid $20 an hour. You can watch some video of the AI cutting wood, managing its inventory, and scouring caves for yourself below. If watching an AI essentially worth the annual salaries of surgeons playing an 11-year-old, indie game isn't exactly impressive It's worth taking a step back and seeing how far technology has advanced. Three years ago teams of technologists participating in the MineRL competition were tasked with one simple task: to create an AI that could successfully mine a diamond in Minecraft. There were 660 contestants who tried to finish the task, and every last one of them failed. OpenAI's model now can create diamond tools. OpenAI isn't the sole tech company that is turning to Minecraft for its AI experiments. Last month, during its Build conference, Microsoft revealed a new AI Minecraft "agent" that is part of the game. When interacting with Microsoft Minecraft agents, users can type commands into the game, which are automatically generated by the game's API. In practice, Wired notes, that means that players can type an expression like "come here" and the Minecraft bot will translate it into Minecraft code, resulting in the robot actually moving forward. In addition to walking in the direction of Microsoft's Minecraft agent can also perform more complex tasks such as retrieving items out in the game world, and combining the items to create something. It can do this faster and better than this writer, who is several years removed from his last Minecraft session.