Creating Cleverer & Cleverer Machines
The smartphone we carry around is an automated learning machine as are large companies like Google and Amazon. While the current stage of artificial intelligence is our de facto librarian, social arbiter, pocket dietitian, and geospatial guide it is also running entire manufacturing plants, NASA missions, financial markets, wars, and more. At the end of the day, these automated systems are learning and making decisions without any human intervention.
According to Computational Linguist and president of the International Machine Learning Society Dr. Pedro Domingos, author of “Master Algorithm” published in February 2015, machine learning is a new paradigm in computing allowing computers to learn to perform important tasks by automating the discovery of existing or better methods of doing things. “Master Algorithm” comes along at a time when a huge explosion of data is making traditional methods of thinking, learning, and discovery impractical and obsolete.
The Automation of Discovery
The toolbox for building learning machines includes: theoretic models, probabilistic logic, linguistic background knowledge to fine tune inferences, the creation of a library of problem-solving algorithms, automated algorithm construction, algorithm function optimization, and much more.
The conclusion in Domingos’ book is that all future knowledge will be derived from “universal learners” that squeeze information out of data. The output of these algorithms includes the creation of new and perhaps more effective learning algorithms.
The following video is a famous TED Talk by Kirk Borne titled “Big Data, Small World” and goes into details and provides examples of how our world is being transformed by learning machines and big data.