DeepMind, owned by Alphabet, which is also owner of Google, is a venture that specializes in Artificial Intelligence (AI). Recently, DeepMind has announced that it has succeeded in developing a computer that can refer to its own memory to learn facts and use that knowledge to answer questions.
We know DeepMind because earlier this year, the company defeated the world’s best GO player with their AI system. Now, its new model, the differentiable neural computer (DNC), has just learned how to navigate in the London Underground. The DNC was able to calculate the fastest route between different Underground stations. It was able to do this by combining an external memory with deep-learning. In deep-learning, the DNC learns to carry out tasks by itself without the input of human-programmed rules.
The system uses the external memory similar to humans using their short-term memory. The makers of the AI told the Guardian that they are wary of saying that they have developed a machine that can reason, like a human. The achievements DeepMind has made change the current view on AI. In the future, it will become easier to apply deep learning to other, more complicated tasks. Even tasks that were always thought to be too difficult for an AI system.
With systems becoming more and more human-like, the science-fiction component in Artificial Intelligence is disappearing. We already have self-driving cars driving around for years. However, especially in the past year AI-technology has made big steps. Not only Google is experimenting with AI, but Microsoft and Apple are experimenting with AI as well, not to mention the hundreds of small tech start-ups. Some even say that an AI-technology revolution is on its way. Apple’s CEO, Tim Cook, for instance, is one of the leading industry experts who think artificial intelligence is soon to conquer the world.
Sources:
http://www.nu.nl/internet/4335611/google-dochter-wil-computer-indirecte-verbanden-laten-leggen-.html
http://www.thedrum.com/opinion/2016/10/07/the-ai-ad-tech-revolution-the-way
