The limits of artificial intelligence (AI) research and implementation are being pushed at British tech company, DeepMind.

The innovative and futuristic nature of this work was signalled by October's creation of an AI ethics committee by the company to help prepare the public for the wider impact of AI.

[Read more: What is Artificial Intelligence?]

What is DeepMind?

DeepMind is a British AI research company. It was founded in 2010 and bought by Google in 2014 for £360 million.

Its founders are Dr Demis Hassabis, Mustafa Suleyman and Dr Shane Legg, who last week were also awarded the accolade of topping Business Insider UK’s ‘The 100 coolest people in UK tech’.

DeepMind Founders

Where is DeepMind based?

The company retains space in Google’s London headquarters, and also has an additional research centre in Edmonton, Canada, and a DeepMind Applied team in Mountain View, California.

What does DeepMind do?

“We’re on a scientific mission to push the boundaries of AI, developing programs that can learn to solve any complex problem without needing to be taught how,” reads the grand ambitions of the company from its website.

The company believes that AI will be able to help in sectors as diverse as climate change and healthcare.

[Read more: Google's DeepMind to start playing StarCraft II as part of machine learning research]

DeepMind and AlphaGo

Go boardgame

DeepMind’s most famous research was when it created an AI programme that beat the world’s best player at Go, a complex logic-based board game.

The game, watched by over 200 million people across the globe, happened in Seoul, South Korea against world champion Lee Sedol.

[Read more: Google's DeepMind AI beats Go world champion in first match]

Even though the computer beat Lee 4-1 over a week-long showdown, he said he did not believe AlphaGo had surpassed human intelligence.

"I don't necessarily think AlphaGo is superior to me. I believe that there is still more a human being could do to play against artificial intelligence,” Mr Lee said.

"When it comes to psychological factors and strong concentration power, humans cannot be a match.”

The method behind created AlphaGo was published in the scientific journal, Nature.

DeepMind and AlphaGo Zero 

On 18 October, DeepMind revealed that it had created a newly version of AlphaGo that defeated its previous iteration 100 games to zero after just three days of training. This latest evolution of AlphaGo was trained not against human games but learnt against itself in self training. 

AlphaGo Zero

 

(Gif from DeepMind) 

DeepMind and chess

In December, a non-peer-reviewed research paper published with Cornell University Library’s arXiv, found that DeepMind's AlphaGo had taught itself to beat the world champion of chess in just four hours. 

DeepMind AI teaching itself to walk

In July, DeepMind created a program capable of teaching itself to walk and jump without prior input, which led to some rather unusual results.

A paper on the work published in Cornell University Library states that the technology uses the reinforcement learning paradigm – which allows “complex behaviours to be learned directly from simple reward signals”.

The avatars traversing the habitats included a humanoid, a pair of legs and a spider-like four-legged creation tasked with leaping across gaps.

[Read more: Google’s DeepMind AI has taught itself to walk]

DeepMind and the NHS

DeepMind Health is a branch of DeepMind that deals specifically with its work in the NHS, and was launched in February 2016.

However, it’s not all been plain sailing. Its trial with a London hospital has led to the nation’s information rights regulator saying it did not comply with data protection law, and a public admittance by the company of the need to be more open with its move into the healthcare system.

DeepMind has currently got its acute kidney injury alert app Streams, or plans to install it, at three NHS hospitals, and has research partnerships with two trusts.

[Read more: NHS to use Google's artificial intelligence technology]

AI and ethics

On October 3, 2017, DeepMind announced a launch of its Ethics & Society research unit. The company says the group will explore and understand the real-world impacts of AI, and aims to help technologists put ethics into practice and help society learn more about the impact of AI.

It is being led by Verity Harding and Sean Legassick, who said in a blog post on the launch day: “We believe AI can be of extraordinary benefit to the world, but only if held to the highest ethical standards”.

[Read more: Google's new website can guess what you are doodling]