How Deep Learning Systems Are Beginning to Surpass Humans
From the stuff of dystopian science fiction movies to everyday companions – with the rise of ubiquitous mobile computing power, artificial intelligence (AI) is already permeating modern life. As of 2017, deep learning algorithms power our phones’ voice-assistants, recommend the latest movies, and optimise our bike ride to work. AI has been heralded as the new electricity, soon to be found in almost every piece of technology we produce. To the man who has been described as “the father of modern AI”, this is merely the beginning. Although the artificial neural networks of Jürgen Schmidhuber’s team are now in 3 billion smartphones, he considers our current state of AI technology to be in the early stages of infancy. Whereas today’s seemingly smart algorithms are geared towards singular purposes – playing chess, matching love-hungry 30-somethings, or finding appropriate music for cooking – Jürgen’s goal has always been to create a general-purpose AI within his lifetime. His entire career has been dedicated to developing a software that would outsmart him, and though he readily admits that, as of now, the best general-purpose AI is only comparable to the intelligence of an infant animal, he is convinced that it will not be long before we develop systems that are far superior to us. At Falling Walls, Jürgen lays out the state of the art in his field of research and shares his vision of a future in which humans are no longer the crown of creation.