Today’s AI still miles off equalling human common sense reasoning.
Levesque makes a key point about the direction of current AI research. Such research focusses on crunching big data, using ever more sophisticated machine learning. It is a wonderful tool for a certain set of challenges or problems. But it does not begin to emulate a human’s ability to reason in novel circumstances. Levesque argues that humans benefit from building up an extraordinary bank of knowledge, which underpins common sense reasoning. Human language enables the transfer of this type of knowledge into human heads. We still do not know, however, how the human brain manipulates such knowledge to deliver common sense reasoning. As such, we still do not know how to design an AI to do the same thing.
Link to book: https://www.amazon.co.uk/Common-Sense-Turing-Quest-Press/dp/0262535203/ref=sr_1_fkmr0_1?s=books&ie=UTF8&qid=1522144447&sr=1-1-fkmr0&keywords=Common+sense%2C+the+Turing+Test%2C+and+the+quest+for+real+AI+by+Hector+J+Levesque+2017
You may also like to browse other AI book posts: https://www.thesentientrobot.com/category/ai/ai-books/