DeepMind

DeepMind

Part of a series on AI / Artificial Intelligences. [View Related Entries]

Updated Jan 25, 2019 at 05:37PM EST by Matt.

Added Oct 19, 2017 at 02:24PM EDT by Don.

PROTIP: Press 'i' to view the image gallery, 'v' to view the video gallery, or 'r' to view a random entry.

This submission is currently being researched & evaluated!

You can help confirm this entry by contributing facts, media, and other evidence of notability and mutation.

About

DeepMind is a neural network created by Google's DeepMind Technologies Limited artificial intelligence company, which can play various games, program animated figures and mimic certain functions of the human brain.

History

On September 23rd, 2010, DeepMind Technologies Limited was founded by Demis Hassbis, Shane Legg and Mustafa Suleyman. In January 2014, the company was acquired by Google for $500 million, leading the company to create an artificial intelligence ethics board, though the members of the board have not been revealed.[1] That year, the Cambridge Computer Laboratory awarded DeepMind "Company of the Year."

Highlights

Video Games

On March 7th, 2015, the Two Minute Papers YouTube channel uploaded footage of DeepMind learning how to play the 1976 arcade game Atari Breakout (shown below).



StarCraft II Win

On January 24th, 2019, DeepMind scientists Oriol Vinyals tweeted that DeepMind had defeated two professional StarCraft II players. He tweeted, "Happy that we could share #AlphaStar progress with you all! Good Games @LiquidTLO and @Liquid_MaNa, and @Artosis and @RotterdaM08 for a great show! You can see all the details in the blog." The tweet received more than 420 retweets and 1,300 likes in 24 hours (shown below).




In the accompanying blog post,[6] they wrote:

We introduce our StarCraft II program AlphaStar, the first Artificial Intelligence to defeat a top professional player. In a series of test matches held on 19 December, AlphaStar decisively beat Team Liquid’s Grzegorz "MaNa" Komincz, one of the world’s strongest professional StarCraft players, 5-0, following a successful benchmark match against his team-mate Dario “TLO” Wünsch. The matches took place under professional match conditions on a competitive ladder map and without any game restrictions.

The post concludes that the machine won because of "superior macro and micro-strategic decision-making, rather than superior click-rate, faster reaction times, or the raw interface." These include making more human-like actions, relying less on speed and more on strategy.[7]

AlphaGo

In October 2015, the DeepMind-powered AlphaGo program, used to play the Chinese strategy board game Go, beat the European Go champion Fan Hui, marking the first time a professional Go player had been beaten by a computer opponent. On May 1st 2016, YouTuber ColdFusion uploaded an episode about AlphaGo's victory (shown below). Over the next two years, the video gained over two million views and 4,900 comments.



On October 18th, 2017, the DeepMind YouTube channel released a video on AlphaGo Zero, a new version of the program which was able to learn how to play Go from scratch without given data from human players (shown below).



Walk Animation

On July 12th, 2017, the Tech Insider YouTube channel uploaded a video showing various animated figures created by DeepMind and used to teach itself how to learn to walk and traverse obstacles (shown below). Within three months, the video received upwards of 4.3 million views and 5,000 comments.



Controversies

DeepMind Health

On April 29th, 2016, New Scientist[4] reported that DeepMind had obtained a data-sharing agreement with the Royal Free London NHS Foundation Trust, noting that DeepMind had been given access to sensitive medical information without patient consent.

On July 3rd 2017, Bloomberg and CNBC, as well as other media outlets, outlined how Deepmind's trial app had violated UK law over patient's data. DeepMind admitting partial wrong-doing, but also deflected and offered several explanations as to why it hadn't broken any laws, only skirted the rules.[4][5]

Search Interest

External References

Recent Videos 2 total

Recent Images

There are no images currently available.


Top Comments

Nedhitis
Nedhitis

That is neat from a science perspective, but from a gaming perspective, that is rather cheap, no..? For those that might relate: do you remember Age of Empires II's recording feature..? It allowed you to see the CPU's point of view and all their actions, and man, that was absolutely insane.

The CPU has no concept of camera, clicking or even the map's "fog of war" (unexplored parts of the map). They just automatically do all the actions it needs via scripts and reacts automatically to everything happening in the game, even your own actions they should have no way of knowing you ever did in the first place. The CPU knows exactly where all your units and buildings are the second they are moved/placed. And indeed, if you look at the Twitter video, that is exactly what happened here. Just look how spastic the CPU is moving around and selecting things seamlessly at blink speed.

So yes, no hard feelings for those defeated by this AI. Normal CPUs are already unfair enough: imagine that with an actual pseudo-thinker on top of it. Not even gamers can rise up to this challenge.

+22

+ Add a Comment

Comments (31)


Display Comments

Add a Comment


O HAI! You must login or signup first!