Deep Thinking: Where Machine Intelligence Ends and Human Creativity Begins

by Garry Kasparov

Google doesn’t even worry very much about hiring people with language skills. They feed the system examples of correct translations, millions and millions of examples, so the machine can figure out what’s likely to be right when it encounters something new. When Michie and others tried this in the early days, their machines were too slow and their data collection and entry systems were paltry. No one could imagine that solving such a “human” problem like language could be a matter of scale and speed. They were like the early chess programmers looking at Type A brute force programs and despairing they would ever be fast enough to play competently. As one Google Translate engineer put it, “When you go from 10,000 training examples to 10 billion training examples, it all starts to work. Data trumps everything.”