Gamasutra is part of the Informa Tech Division of Informa PLC

This site is operated by a business or businesses owned by Informa PLC and all copyright resides with them. Informa PLC's registered office is 5 Howick Place, London SW1P 1WG. Registered in England and Wales. Number 8860726.


Gamasutra: The Art & Business of Making Gamesspacer
View All     RSS
June 17, 2019
arrowPress Releases








If you enjoy reading this site, you might also want to check out these UBM Tech sites:


New DeepMind AI learns to play Atari 2600 games at human levels

February 26, 2015 | By Alex Wawro




"We demonstrate that the deep Q-network agent, receiving only the pixels and the game score as inputs, was able to surpass the performance of all previous algorithms and achieve a level comparable to that of a professional human games tester."

- An excerpt of a study on AI that teaches itself to succeed at complex tasks (like playing Atari games 2600 games) published in Nature.

AI-minded developers, take note: A team of researchers at Google DeepMind claims to have developed an articial agent capable of learning to play (from scratch) 49 different Atari 2600 games at skill levels comparable to human beings.

Evidence to back up their claims is laid out in a study published in Nature this month, a full copy of which is available for purchase via the Nature website. You can also see footage of the AI agent, known as "DQN", playing games like Space Invaders and Breakout in a Nature video (embedded above) on the topic.

In brief, the researchers claim to have built an autonomous agent that, when tasked to maximize its score in a title like Breakout with no prior understanding of the rules of the game, learns to play effectively (developing tricks like knocking the ball back behind the brick layer) based on a combination of deep learning and reinforcement learning techniques.

The researchers were interested in developing artificial agents that can successfully learn complex tasks from scratch, not ace old Atari games, but their work is likely to impact the future of game development.

It's also worth noting that the research group includes Demis Hassabis, the former Bullfrog developer who (at the age of 17) co-designed and programmed Theme Park with Peter Molyneux. He later went on to found AI tech developer DeepMind in 2011, which was then acquired by Google early in 2014.



Related Jobs

Gear Inc.
Gear Inc. — Hanoi, Vietnam
[06.15.19]

Technical Director
Legends of Learning
Legends of Learning — Washington, DC, District of Columbia, United States
[06.14.19]

Senior Unity Engineer - $140k - Remote OK
Wargaming.net
Wargaming.net — Chicago, Illinois, United States
[06.14.19]

Server Engineer
Wargaming.net
Wargaming.net — Bellevue , Washington, United States
[06.14.19]

UI Engineer









Loading Comments

loader image