The whole idea behind products, such as Siri, is to give computers a friendlier face. Much like the computer on the Enterprise in Star Trek, you converse with the machine and get intelligent answers back much of the time. The problem is that computers don’t currently have common sense. A computer really doesn’t understand anything anyone says to it. What you’re seeing is incredibly complex and clever programming. The understanding is in the math behind the programming. Computers truly are machines that perform math-related tasks with extreme speed and perfection.
It was with great interest that I recently read an article on the Guardian, Google a step closer to developing machines with human-like intelligence. The opening statement is misleading and meant to bedazzle the audience, but then the article gets into the actual process behind computers that could emulate common sense well enough that we’d anthropomorphize them even more than we do now. If the efforts of Professor Geoff Hinton and others are successful, computers could potentially pass the Turing Test in a big way. In short, it would become hard to tell a computer apart from a human. We very well could treat them as friends sometime in the future (some people are almost there now).
Articles often allude to scientific principles, but don’t really explain them. The principle at play in this case is the use of sentiment analysis based on words and word n-grams. You can build a sentiment analysis by using machine learning and multiclass predictors. Fortunately, you don’t have to drive yourself nuts trying to understand the basis for the code you find online. Luca and I wrote Python for Data Science for Dummies to make it easier to understand the science behind the magic that modern applications seemingly ply. Let me know your thoughts about the future of computers with common sense at [email protected].
