When thinking about the future of AI, it is important to take stock of where we are. Throughout the course of this book, we have tried to make it clear that the vast majority of the capabilities in AI today are centered on recognition and classification tasks. That, in itself, is profound—representing the first thing any human does in any situation that requires reasoning.
However, just because you can recognize something and classify it doesn't mean that you truly understand its nature, its mechanics, its potential, or any number of other things that are important when reasoning about that thing. Likewise, recognition is a far cry from being able to reason deeply about a subject, or create new plans for tackling a problem, or imagining something that has never been created before. Additionally, none of these equates to the deeper motivations of emotion, such as guilt, happiness, frustration, anger, or love. All of these are aspects of human intelligence that we take for...