Forbes – Kalev Leetaru
“Today’s deep learning solutions truly are magical, able to reduce the complexities of the world around us into correlative equations and encode the patterns of our lives into software. Watching a deep learning system at work for the first time is almost like witnessing a magician on stage as piles of training data are analyzed by an algorithm that eventually returns a mathematical summary of the underlying patterns. It is mesmerizing to watch as the pandemonium of petabytes is condensed by machines into the precision of mathematical equations. To watch chaos reduced to clarity. At the same time, these primitive correlative engines can be readily led astray even with the most carefully curated data, leading them to encode biases and invalid patterns.” (more)