Aristotle -Deep Learning-Risk

A few weeks ago, I saw a picture of Aristotle. The picture of Aristotle brought up memories of me in high school trying to understand what philosophers in his generation were debating. I didn’t understand it all. For me the world was simple. Work hard and life gets better.

A key tenet for Aristotle was “why” – why does something happen, why does something move, why does something have form, why is something unique, etc. Simple three letter word. Why – a word most often never uttered by humans.

Time to regroup. I needed to understand why I am where I am, why the world is in the state that it is, why the world will be in the state that it is in thirty days, a year from now, etc. Big task. Five hundred, a thousand pages, and I might scratch the surface. Time to refocus.

Laser beam focus. Around the globe, there are folks that ask why. They have a need to understand “why” something happens the way that it does. An engineer at an automobile manufacturer needs to understand why an engine functions the way that it does. A physician needs to understand why a virus spreads the way that it does, a master carpenter needs to understand why a piece of furniture has both function and beauty.

On one side there are the folks asking why and on the other side there are folks accepting and doing what they are told to do. When most folks step on the gas pedal, they expect the car to move. They don’t ask why. When they buy something from Amazon they don’t ask why they are buying the item. They want it and they buy it. When folks are told to dig a ditch, they don’t ask why the ditch should be dug. They dig.

Over the years, I have come to realize that those that ask why and those that simply respond and do are equally important for a society to succeed. But there is also a chain of command. Those that ask why and understand , give direction to those that implement. Those that implement have faith that those that direct, understand, and are making the correct choices – they understand why something is working the way that it is. There is trust.

Very early in my career, I had just finished taking a deep dive into parallel processing. I had worked with fuzzy logic, genetic algorithms, and rule based systems. I came across some images of neural nets. I immediately started to associate the neural nets with parallel processing of knowledge based upon experiences. I lucked out and was able to take a trip to Pittsburgh and spend some time with folks involved in a start up from Carnegie Mellon. My take away was new insight, models and a stack of manuals. Incredible that digits could be classified:)

Collecting the data — collection of experiences — training, and the conclusion closely resembled something that a human would propose. Breath taking.

Fast forward, and deep learning soon became a component in all sorts of systems. To successfully market and sell, “we are AI based” has become the hook. But along with the rapid acceptance came the result that something like ninety percent of AI based solutions failed. It is a given that the lack of MLOps is part of the reason that so many AI based systems fail. However, really cool solutions but ultimately of little value and even worse dangerous — humans take the results of AI reasoning as they would take direction from a human that asked why and reasoned why one choice was the correct choice out of many.

Far too many implementations of AI based systems are validated upon a criteria that if the results are equal or better than those of a set of humans, then the model works. This completely ignores what Aristotle wrote over two thousand years ago. There are those that ask why and understand. There are those that have experiences and act upon those experiences without asking why. Somewhere along the line, the message was lost and the consequences have proven to be very bad. Look at the state of the world. Don’t simply accept. Reason and understand why before you accept.