Lazy deep learning

The neural net, an old technology, improved recently by marginal tricks for faster and stable learning, is now the most expensive old wine in a new bottle, appropriately called, "deep learning." Technologists have been getting more creative and now they believe they are going to take over the world. It is possible, but unlikely. Scientists, who have been struggling with age-old statistics suddenly find a way to throw large amounts of raw data to the dumb machine to make pattern finding easier. We could now chase fundamental particles, pharmaceutical products, and weather forecasts, by chasing noise. Now, one could automate it with insights rising to the top just like butter does as one churns spoiled milk.

Lack of experience is problematic; for politicians, management consultants, investment bankers, entrepreneurs, scientists and even bureaucrats. It takes a while to be good at what one does. Machines are certainly great but as a large mainframe maker found out recently down South that not even Sherlock, let alone Watson, can solve all the world's problems. And the largest and smallest analytics companies in the world clamoring for glory by the application of technology and "artificial intelligence," some to save humanity and others to stuff their own pockets, we are fast approaching a highly bifurcated regime. As they seek "deep mind," much deeper questions remain and that's not something the millennials appear to be interested in.

Preserve the human mind, compassion and an incessant yearning for knowledge to survive. Sometimes, it is better to take a break from "learning programmatically."