From Wikipedia: Early mentions and definitions of intuition can be traced back to Plato. In his book Republic he tries to define intuition as a fundamental capacity of human reason to comprehend the true nature of reality. In his works Meno and Phaedo, describes intuition as a pre-existing knowledge residing in the “soul of eternity”, and a phenomenon by which one becomes conscious of pre-existing knowledge. He provides an example of mathematical truths, and posits that they are not arrived at by reason. He argues these truths are accessed using knowledge already present in a dormant form and accessible to our intuitive capacity. This concept by Plato is also sometimes referred to as anamnesis.
Intuition, as a gut feeling based on experience, has been found to be useful for business leaders for making judgement about people, culture and strategy. Law enforcement officers often claim to observe suspects and immediately “know” that they possess a weapon or illicit narcotic substances, which could also be action of instincts. Often unable to articulate why they reacted or what prompted them at the time of the event, they sometimes retrospectively plot their actions based upon what had been clear and present danger signals. Such examples liken intuition to “gut feelings” and when viable illustrate preconscious activity.
How can you put this into a computer program? How can intuition be coded if it is not easy to even define it? What could the underlying math and theory be? How do you formulate gut feeling in mathematical terms? Is it at all possible? Will computers ever exhibit common sense (which, by the way, amongst humans appears to be the least common of the senses)?
And what would the advantages of a piece of computer code be if it really could exhibit intuition or gut feeling? How about the following:
- Speed. Intuition is immediate. This is because it is instinctive.
- No need to generate thousands of relevant examples to learn from.
- No need to update the learning set when things change even not so dramatically.
- Intuition is ‘universal’ – you don’t need to ‘tune it’ to a particular field of application, you just have it.
One of the properties of intuition is that it is not precise. This, however, is not at all a problem. In fact, intuition is most valued in highly complex circumstances in which:
- it is impossible to make predictions
- only very rough estimates can be attempted
- there is no such thing as precision
- it is impossible to isolate “cause-effect” rules as everything is linked
- optimization is unjustified – one should seek acceptable solutions, not pursue perfection
- there are very few examples to learn from
The well known Principle of Incompatibility states in fact that “high precision is incompatible with high complexity”. This means that in conditions of high complexity:
Precise statements are less relevant
Relevant statements are less precise
However, this fundamental principle, which applies to all facets of life, as well as to Nature, goes unnoticed. Neglecting the Principle of Incompatibility clashes with reality and hard facts. Already Aristotle wrote in his Nikomachean Ethics: an educated mind is distinguished by the fact that it is content with that degree of accuracy which the nature of things permits, and by the fact that it does not seek exactness where only approximation is possible. The bottom line is that if something is highly complex it won’t be precise. You cannot have both.
So, it appears that Artificial Intuition is most useful in scenarios that, due to the very nature of things, are fuzzy, imprecise, non-stationary and complex. In such contexts running analyses that provide precise answers is senseless.
QCM (Quantitative Complexity Management), which distinguishes Ontonix, is one way of doing Computational Intuition, or Artificial Intuition. Clearly, the use we make of this technology is to quickly identify:
- what is really important
- sources of fragility, vulnerability
- malfunctions, ruptures
- imminent crises
So how does all this stuff work in practice? Experience suggests that when things get complicated, when they run out of hand, or one loses control, there will be problems. It is precisely the ability to identify that the complexity of a given situation or system is rapidly rising provides the intuitive component. Evidently, there is no need to learn. All it takes is monitoring. An example of a system suffering three consecutive crises is shown below:
The peaks in the complexity function correspond to crises, while the rapid increase in slope provides the predictive component. No examples are necessary, no need to learn, no previous knowledge. Elementary Watson!
QCM is AI, Artificial Intuition.
Pingback: Artificial Intuition Finds Unknown Anomalies – Beyond Machine Learning – Quantitative Complexity Management