What Constitutes Artificial Intelligence? Is It the Turing Test?

Exploring the use cases for 'AI' nowadays is more popular than defining it, I would like to see more dialog around discussing what constitutes 'Artificial Intelligence'.

As with most of us in this field, I hate the often-overused term. More aptly, I believe these are 'Decision Making Machines', they really don't do Critical Thinking, or Abstract Thinking, they simply have a set of given inputs, and a function that decides which set of those given inputs matches a given output.

There's no better example of this than Watson, and Medical Diagnosis. That is not AI. We have an enormous database full of years of medical history. From that history, we draw conclusions about symptoms. For a patient's given symptoms, we are often able to crawl back up that ladder to the diagnosis. Yet we still call Watson AI.

Sometimes, these decisions are chained, giving the illusion of Critical Thinking, but at the end of the day, they are goal-driven algorithms.