Slashdot Mirror
OpenAI Built Gaming Bots That Can Work As a Team With Inhuman Precision (qz.com)
OpenAI said on Monday that its newest AI bots can hold their own as a team of five against human gamers at Dota 2, a multiplayer game popular in e-sports for its complexity and necessity for teamwork. The AI research lab is looking to take the bots to Dota 2 championship matches in August to compete against the pros. From a report: Dota 2 is a challenging game for AI to master simply because of the amount of decisions that the players have to juggle. While chess can end in fewer than 40 moves, and Go fewer than 150, OpenAI's Dota 2 bots make 20,000 moves over the course of a 45 minute game. While OpenAI showed last year that the bots could go one on one against a human professional in a curated snippet of the game, the company wasn't entirely sure that they could scale up to five against five.
But the research team doesn't credit this breakthrough to a new technique or a lightbulb moment, rather a simple idea. "As long as the AI can explore, it will learn, given enough time," Greg Brockman, OpenAI's chief technology officer, told Quartz. The bots learn from self-play, meaning two bots playing each other and learning from each side's successes and failures. By using a huge stack of 256 graphics processing units (GPUs) with 128,000 processing cores, the researchers were able to speed up the AI's gameplay so that they learned from the equivalent of 180 years of gameplay for every day it trained.
But the research team doesn't credit this breakthrough to a new technique or a lightbulb moment, rather a simple idea. "As long as the AI can explore, it will learn, given enough time," Greg Brockman, OpenAI's chief technology officer, told Quartz. The bots learn from self-play, meaning two bots playing each other and learning from each side's successes and failures. By using a huge stack of 256 graphics processing units (GPUs) with 128,000 processing cores, the researchers were able to speed up the AI's gameplay so that they learned from the equivalent of 180 years of gameplay for every day it trained.
So children aren't intelligent, since they need training to do things like read and write, or even how to talk?
File under 'M' for 'Manic ranting'
Why does this conjure up images of hordes of inhumanely fast robots swarming cities and taking out citizens and soldiers with ease... How long until there forms an upperclass completely immune to revolution or the conscience of its human military?
That makes sense, since Quake is indeed an Inhuman.
...it can be applied to important things like X and Y and Z! We promise. It isn't just for games. This is real important stuff.
These are just buzzwords and their meaning is evolving. The latest distinction between machine learning and AI that I saw is that we let AI make decisions based on it's training. In other words, AI is just ML with the handcuffs taken off. Within either, you still have directed or non-directed training.
When humans start looking to bots to figure out the meta game that'll take out half the fun
From TFA: The bots learn from self-play ...
The equivalent of 180 years of self-play must have left them deaf as an adder.
(OK, I admit, I'm really a bot that learned how to post jokes through self-play)
I learned a lot from 'self-play'; does that count?