Heuristic input to Predictive Analytics

We need to find ways to build common sense into artificial intelligence

Cover Illustration for Inventing the Future with Haiku: Whistler Village, computer painting by the author

Because Artificial Intelligence (AI) has the ability to process ‘big data’ it can be applied to huge problems involving complex systems. However, in order to be reliable in forecasting the future, it needs to incorporate the intuitive aspect of human intelligence. How can that be programmed?

It is possible that properly coded ­algorithms might eventually enable a computer to execute heuristic processes (roughly defined as common sense), but it is more likely that heuristic processes can serve as a good first step in data analytics by synthesizing data into a form AI can handle.

Glean as much as you can intuitively before you start quantifying and fitting number-crunching formulas.

I figured this out almost fifty years ago while studying what is now called business analytics. It was called management science when I got my MBA, and operations research before that. There was probably a lot less data to mine in the early `70s, but it seemed like a lot at the time.

1. The Delphi Technique:

This is a method of arriving at a group opinion or decision by surveying a group of experts (usually in diverse fields) who respond anonymously, then have an opportunity to reassess their answers after…