We know it as a decision support issue, but when he was United States Secretary of Defense, Donald Rumsfeld brought some criticism on himself with his talk about ‘known unknowns’. The pronouncement quickly passed into journalistic lore, but there was much sense in what he said. Unfortunately, his memorable, confusing and apparently paradoxical phrasing of the issue has devalued its importance in the mind of many people.
When the average human being makes a decision, it is based on information and a knowledge of the scope of that information in the context of the known world (to the extent that it is understood by that individual). There is also an element of wisdom – that is, a weighting of the information and its contextual position based on the memory of similar situations and outcomes.
A computer may be programmed to make a decision based on data, and even include historical outcomes to weight the factors in the decision based on the quality of that data. The imperfect nature of data can recognized (if recorded) and can be handled programmatically. The computer analysis of a situation cannot include factors for which it has not been programmed, or data of which it is unaware.
The fact is that the general solution has been written into the computer program by a human, though the program may even be able to rewrite itself and improve, by means of feedback.
A human being can transform data into information and store it as knowledge – recognizing and using it as such, and using it as a weighted component for decision support.
What computers cannot at present do as well as human beings is to deal with new situations as events unfold in the real world.
Parallels and similarities can be hard to determine – one only has to look at Captcha codes on data input screens – ‘to show that you are not a robot’.
This is the essence of memory-based reasoning.