Do it Yourself – Role BI at Machine Learning & AI

In the series of articles on the topic of Oracle Business Intelligence, a lot of ground has already been covered. And much more lies ahead. This blog is about automated decisions. In the rapidly changing world Oracle Business Intelligence starts playing an important role introducing features such as Machine Learning and Artificial Intelligence.

If Oracle BI can alert exceptions or deviations or errors and we humans react with a certain repeatable set of actions why shouldn’t we let the computer do that for us?

Introduction

In a previous article about Big Data I mentioned the development of the chess computer. In fact the development of software that has cracked games such as chess and go, is basically foundation of Artificial Intelligence. Based on rules, historical data and brilliant algorithms, a computer can make better and faster decisions for us that work to our advantage. Once this set of conditions is good enough, a computer can draw conclusions better than a human being. In the present time we can already see the first applications of software in health care, already able to diagnose better than the most experienced physician.

Frightening? I do not think so. How beautiful is it? Previously we drove horses. In a few years we will be reading a book in a self-driving car (without a drivers license). This evolution is unstoppable. It’s coming.

Today a chemist looks at a blood sample through his microscope and sees whether something is wrong. Soon a computer will do this for us and if fed with a lot of data, history, rules and science it will do a better job than any human.

If we take one step further, in the (near) future software will be developed by software and robots will be constructed by robots. And I think I can certainly state that this will improve this drastically as both will be much smarter than us humans.

Frightening? I do not think so. Why should we not be assisted, guided and improved by the most optimal predictions and decisions made for us. As long as it is clear and understandable, this is fine. However there comes a time when it is no longer comprehensible for simple souls like us. But which of us can still ride a horse and a car? And who really knows how a refrigerator does what it does?

Machine Learning (ML) or Artificial Intelligence (AI)

Some time ago I watched the movie The Imitation Game. This is about the British secret service which in the Second War is trying to crack the Enigma code in order to decipher German messages. This would take people thousands of years of trial and error. The ancestor of the computer was built to get the job done. But only when this machine is fed with some ground rules and base principles it was able to accomplish this task in a reasonable time.

The current versions of Oracle Business Intelligence show the button Machine Learning. What is that? In a nice introduction of this concept by a venerable employee of Oracle The Netherlands we get a picture.

In a nutshell we help the computer with its massive computing power to first learn controlled. We give the machine a huge amount of data and then we tell it what kind of outcomes we want and can expect. For example: We let the computer take a look at a blood sample which houses traces of a decease. We help the computer conclude the same. This controlled stage provides us with sufficient rules to leave the computer working independently. Then we let the computer take a look at that same blood sample to attend us there is also something else to see.

For years beautiful movies about Artificial Intelligence have been produced. For fans (such as myself) I recommend a few: Delete (2013), Ex Machina (2014), The Machine (2013) and Transcendence (2014). Each and every one of these wonderful forecasting movies are actually the contemporary books by Jules Verne. It is not possible yet, but it can go that way.

Frightening? I do not think so, as long as we keep a close eye on who the most important stakeholder is.

In a beautiful lecture by doctor Asimov he touches this subject. We will always have to tell robots and computers to leave people alone. He means the following: If we let the machines go their own way, let robots build better robots and software to develop better software, they will soon draw a conclusion that threatens our survival: There is only one true danger threatening this planet and that is the human being … But as soon as we limit a condition in robots and software to keep us alive, both will get human-like, meaning they will also be able to decide to start a war …

Frightening? I do not think so, as long as we are very aware of this.

Now what?

Let’s look at the current possible applications on a dashboard in Oracle Business Intelligence that shows us some exceptions, such as a couple of purchase orders not yet received. If the employee in the organization invariably contacts the suppliers, could we then also automate this? In the continuum of Oracle BI this is called Automated Decisions. The rules and logic and actions that we apply based on information are automated. To simplify this we can compare a thermostat in your living room. If it indicates a temperature of 14 degrees in the winter, then we can decide to put it at 21 degrees and it gets warmer in the house. But the modern thermostat can do this automatically. We set a comfortable temperature and the thermostat decides to activate the heater when necessary. If we take this into the future, we can provide this device with centuries of historical meteorological data and smart predictive power and the device can preadjust for the weather the day after tomorrow. Frightening? No. Wonderfully comfortable and easy. That’s the way it should be with Artificial Intelligence. That is true added value for our lives with devices that are much faster and much better and more precise than us humans beings at drawing the right conclusions and taking the correct actions.

Recently I have heard about a prediction that it is feasible in the future that the self-driving will make zero traffic casualties feasible. World-wide that is…

Curious for more? Soon this will have a follow-up!

Author:  Rick Brobbel
BI Consultant at Cadran Consultancy