Microsoft has announced a new partnership with OpenAI, the $1 billion artificial intelligence research nonprofit cofounded by Tesla CEO Elon Musk and Y Combinator President Sam Altman. Under the terms of the partnership, OpenAI will use the Microsoft Azure cloud computing service as the "preferred" place to do its AI experiments. In return, Microsoft gets increased access to OpenAI's deep bench of robotics and experts, making Azure a better place for building AI-powered software. "You want to get a good feedback loop going," says Microsoft Executive VP of Cloud and Enterprise Scott Guthrie. "That helps you ultimately build a better platform."

For millennia, we surpassed the other intelligent species with which we share our planet--dolphins, porpoises, orangutans, and the like--in almost all skills, bar swimming and tree-climbing. In recent years, though, our species has created new forms of intelligence, able to outperform us in other ways. One of the most famous of these artificial intelligences (AIs) is AlphaGo, developed by Deepmind. In just a few years, it has learned to play the 4,000-year-old strategy game, Go, beating two of the world's strongest players. Other software developed by Deepmind has learned to play classic eight-bit video games, notably Breakout, in which players must use a bat to hit a ball at a wall, knocking bricks out of it.

Google's AI achieved record scores playing Atari and beat a human during a game of Go – but to do so, the system needed constant training over a long period of time. In order to speed up and simplify the learning process, the team at DeepMind has now taught its AI to dream in a way that is similar to animals. After replaying sequences of a game containing rewarding events, the system proved to learn the stages 10 times faster than previous algorithms. To speed up the learning process, the team at DeepMind has now taught its AI to dream in a way that is similar to animals. After the system began'dreaming' about a game called Labyrinth (pictured), it proved to learn the stages 10x faster than the previous algorithms DeepMind's system Unreal was augmented with two additional tasks.

When it comes to machine learning, every performance gain is worth a bit of celebration. That's particularly true for Google's DeepMind division, which has already proven itself by beating a Go world champion, mimicking human speech and cutting down their server power bills. Now, the team has unveiled new "reinforcement learning" methods to speed up how the AI platform trains itself without being directly taught. First off, DeepMind's learning agent has a better grasp of controlling pixels on the screen. Google notes it's "similar to how a baby might learn to control their hands by moving them and observing the movements."

But the newest artificial intelligence system from Google's DeepMind division does indeed dream, metaphorically at least, about finding apples in a maze. Researchers at DeepMind wrote in a paper published online Thursday that they had achieved a leap in the speed and performance of a machine learning system. It was accomplished by, among other things, imbuing technology with attributes that function in a way similar to how animals are thought to dream. The paper explains how DeepMind's new system -- named Unsupervised Reinforcement and Auxiliary Learning agent, or Unreal -- learned to master a three-dimensional maze game called Labyrinth 10 times faster than the existing best AI software. It can now play the game at 87 per cent the performance of expert human players, the DeepMind researchers said.

Microsoft has announced it will be partnering with the Elon Musk-backed OpenAI initiative in a bid to make "significant contributions to advance the field of AI". As part of its stated mission to'democratise AI', Microsoft will work closely with OpenAI in a joint effort to help ensure artificial intelligence is used to tackle some of the world's most challenging problems. The company outlined the new partnership in a blog post, where it also explained that OpenAI will use its Azure cloud platform, to help advance research and "create new tools and technologies that are only possible with the cloud". OpenAI is a nonprofit artificial intelligence research organization co-founded by Elon Musk, Sam Altman, Greg Brockman and Ilya Sutskever. The organisation will be using Microsoft's Azure N-Series Virtual Machines, which are designed to carry out intensive tasks such as deep learning, simulations, rendering and the training of neural networks.

But the newest artificial intelligence system from Google's DeepMind division does indeed dream, metaphorically at least, about finding apples in a maze. Researchers at DeepMind wrote in a paper published online Thursday that they had achieved a leap in the speed and performance of a machine learning system. It was accomplished by, among other things, imbuing technology with attributes that function in a way similar to how animals are thought to dream. The paper explains how DeepMind's new system -- named Unsupervised Reinforcement and Auxiliary Learning agent, or Unreal -- learned to master a three-dimensional maze game called Labyrinth 10 times faster than the existing best AI software. It can now play the game at 87 percent the performance of expert human players, the DeepMind researchers said.

The artificial intelligence (AI) non-profit OpenAI has agreed to partner with Microsoft to develop "cloud brains" to test its experiments. The organization, which is backed by Elon Musk, has signed an agreement that will allow it to run large-scale experiments using the company's Azure cloud services. OpenAI aims to discover more about deep learning and AI, while Microsoft will use the partnership to create new tools and technologies that use AI. OpenAI was one of the first adopters of Microsoft's Azure N-Series Virtual Machines service that was designed to handle the intense computing workloads that are needed to run simulations and deep learning projects. The service is powered by Nvidia's graphics chips and will be made generally available starting in December.

Standouts included Hike, the Indian messaging app that works offline (useful in a country where connectivity is patchy and data is bought in packages) and transcends the dozens of languages and avoids complex keyboards by using digital stickers as tools of communication. Mustafa Suleyman co-founded DeepMind, now owned by Google, and is forging ahead with the application of AI to solve some of the worlds biggest problems. The use of AI diagnosis in medical imaging can speed up treatment times for cancer and improve patient prognosis. DeepMind are attempting to solve the problem of most NHS data currently being written on paper, and therefore largely inaccessible. Mustafa says "In life, data is pushed to us.

The AI that learns by PLAYING: Google DeepMind is making machines'feel' their way around virtual objects Researchers created virtual environment for training artificial intelligence AI learned through a series of experiments where it'played' with objects By manipulating virtual building blocks in a simulation it could work out hidden properties, learning like a child AI learned through a series of experiments where it'played' with objects Google's DeepMind is teaching machines to learn through play, by exploring objects through simulations to work out their properties The ultimate guide to the biggest supermoon in living... Hundreds more species than we thought may be endangered... Alien hunters say they have... Move over James Bond: World's first commercially available... The ultimate guide to the biggest supermoon in living... Hundreds more species than we thought may be endangered... Alien hunters say they have... Move over James Bond: World's first commercially available... If a child were presented with two blocks painted black, one made of wood and one made of lead, they could work out their basic properties through playing with them. So why can't machines learn the same way? Earlier this week, researchers in Italy launched a new project to develop robots which learn by themselves, using a form of open open-ended machine learning.