PREVIOUS ARTICLENEXT ARTICLE
NEWS
By 3 November 2017 | Categories: news

0

In the escalating arms race between mankind and AI, us meat-bags have a paltry track record, having already lost at Go and Dota 2. Our South Korean brothers though, are keeping the flag flying, recently destroying A.I. bots 4-0 at StarCraft, according to MIT Technology Review.

A big shout-out to professional StarCraft player, Song Byung-gu, who beat out A.I. bots from Australia, Norway and Korea, plus one developed by Facebook’s A.I. research lab.

StarCraft A.I. competitions have been held at the Sejong University of Seoul since 2010, pitting different A.I. against each other. This year was the first where man was matched against machine.

MIT Technology Review notes that A.I. find it difficult to come to grips with StarCraft, since the game requires the use of memory, strategy and planning, all taking place in real-time.

And it shows, since Song destroyed all four bots in less than 27 minutes in total. This despite the fact that the Norway A.I. could complete 19 000 actions per minute, with professional human players only able to clock in the range of a few hundred.

Interestingly, Song notes a big difference in the bots’ approach to the game, seeing that they focussed more on keeping their units alive, without taking any chances by attacking. “The way they managed their units when they defended against my attacks was stunning at some points,” he said.

It seems like we should enjoy our victory while it lasts, since Google’s DeepMind and Blizzard Entertainment (the company responsible for StarCraft) released AI development tools that work with StarCraft II. These makes it easier to train bots, signalling what can only be the end to human dominance of the game. 

USER COMMENTS

Read
Magazine Online
TechSmart.co.za is South Africa's leading magazine for tech product reviews, tech news, videos, tech specs and gadgets.
Start reading now >
Download latest issue

Have Your Say


What emerging technology holds the greatest potential?
Artificial Intelligence (125 votes)
Blockchain (23 votes)
Virtual Reality (16 votes)
High Performance Computing (17 votes)
Machine Learning (23 votes)
Nanotechnology (35 votes)
Computer vision (5 votes)
Edge computing (4 votes)
Autonomous vehicles (128 votes)